Mechatronics Engineering and Electrical Engineering [1 ed.] 1138027197, 9781138027190

Citation preview

MECHATRONICS ENGINEERING AND ELECTRICAL ENGINEERING

This page intentionally left blank

PROCEEDINGS OF THE 2014 INTERNATIONAL CONFERENCE ON MECHATRONICS ENGINEERING AND ELECTRICAL ENGINEERING (CMEEE 2014), SANYA, HAINAN, P.R. CHINA, 17–19 OCTOBER 2014

Mechatronics Engineering and Electrical Engineering

Editor

Ai Sheng Information Science and Engineering Technology Research Association (ISET), Hong Kong, China

CRC Press/Balkema is an imprint of the Taylor & Francis Group, an informa business © 2015 Taylor & Francis Group, London, UK Typeset by V Publishing Solutions Pvt Ltd., Chennai, India Printed and bound in Great Britain by Antony Rowe (A CPI-group Company), Chippenham, Wiltshire All rights reserved. No part of this publication or the information contained herein may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, by photocopying, recording or otherwise, without written prior permission from the publisher. Although all care is taken to ensure integrity and the quality of this publication and the information herein, no responsibility is assumed by the publishers nor the author for any damage to the property or persons as a result of operation or use of this publication and/or the information contained herein. Published by: CRC Press/Balkema P.O. Box 11320, 2301 EH Leiden, The Netherlands e-mail: [email protected] www.crcpress.com – www.taylorandfrancis.com ISBN: 978-1-138-02719-0 (Hbk) ISBN: 978-1-315-73446-0 (eBook PDF)

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Table of contents

Preface

xi

CMEEE 2014 committee

xiii

Battlefield situation cognition system based on big electronic reconnaissance data processing S.M. Xu, H.B. Yu & F.H. Chu

1

Automatic recognition of waste rock in top coal caving based on digital image processing B.P. Wang, Y.J. Wang & Z.C. Wang

5

The application of Dijkstra algorithm based on multi-weighted distribution in GIS Sh.M. Wang, H.T. Luo & B.W. Wang

9

Design of battery charging and discharging circuit H.Q. Zhan & Z.P. Li

15

Automatic flight control of a certain UAV using LQG design Sh.Y. Zhang

19

Research on online and remote calibration technology of electronic belt scale based on weights superposition method X.P. Shang, H.Y. Chen & Z.W. Huang

23

The application of wavelet filter in static synchronous compensator L. Zhang & N. Liu

27

Fractional-order differential application and research in pavement cracks image enhancement W. Jiang, Y.W. Liu & H. Zhang

31

Molecular algorithm in solving the shortest path problem of the application and research L. Zhang

37

A method about video quality assessment in video phone service over 3G network C. Feng, L.F. Huang & W.J. Xu

43

Analysis on the degree of the international fragment of China’s manufacturing industry based on processing trade Y.H. Yang

49

The voltage-controlled low-pass filter based on FPGA frequency measurement Sh.Q. Ma, L.R. Zheng, J.F. Liu & Y.P. Ji

53

Departure capacity assessment of close staggered parallel runways J.G. Kong, X. Li & W.B. Ding

59

The control design of blast furnace clay gun play mud quantity B.H. Jiang, J. Mei, X. Zhao, D.C. Gao & H. Liu

63

Research and implementation of coal traffic video management system based on the technology of image tracking and recognition S.J. Jin Residents air conditioning load management model studies and impact analysis B. Li, S.W. Li, S.X. Zhang, X.S. Jing, F. Wu & X.J. Weng

v

67 71

Multi-objective optimized scheduling for hydro-thermal power system W.J. Liu & P.F. Cheng

77

An optimal space-borne Solid-State Recorder based on domestic chips S. Li, Q. Song, J.W. Song, W. Wang, Y. Zhu & J.S. An

83

Study on additional damping controller for VSC-HVDC to prevent low-frequency oscillation Z.H. Wang, Y. Li & X.Y. He

87

Lightning surge analysis for 220 kV AC double circuit transmission line using LPTL Y. Ma, C.S. Liu, Y. Liu, G. Chen, T.X. Xie, Z.C. Zhou & F.B. Tao

91

Fault analysis of Metal Oxide Varistor (MOV) for series compensation capacitor banks Y. Ma, Y. Liu, P. Li, G. Chen & Z.Z. Zhou

97

Experimental study of the human retinal security by LED light L. Zhao, L. Lu, Sh.Y. Ren, L.Q. Geng & C.T. Li

103

Analysis of the phenomenon in LFO of AC electric locomotive J.J. Ding, Y. Tu & S.B. Gao

109

The prototype of fatigue damage detection by the method of Metal Magnetic Memory Testing X.Y. Yu & X.Y. Fu

115

Rotor composite faults diagnosis of asynchronous motor based on complex power Zh. Wang, Zh.F. Lin, Ch. Li & Zh. Zeng

119

Commissioning and operation of Heze some leather wastewater treatment plant K. Wang, B. Liu, H.D. Zhang & W. Zhao

125

Engineering depth treatment of pharmaceutical wastewater by magnetization catalytic oxidation-A2O-MBR combined process W. Zhao, B. Liu, H.D. Zhuang, K. Wang & H.D. Ji Study of broadband microstrip circulator using YIG single crystal Zh.Q. Cheng, Y. Luan, M.Sh. Jia & X.X. Lian

129 133

Decoupling mechanism for suspension laryngoscopy using a curved-frame trans-oral robotic system X.Y. Fu, X.Y. Yu & J.T. Seo

137

Calculation and analysis on shunt coefficient of short-circuit current inside Fuzhou 1000 kV Ultra-High Voltage substation B. Zhang, B. Tang, W.H. Ma & J.Y. Zou

141

Mechanical analysis of reliability test device of Head-Up Display’s retraction and extension Y.G. Liu & Y.X. Wang

147

Study on electromagnetic loop rejection in Liaoning power grid K. Gao & Zh.H. Wang

151

Study on effect of evaluating corrosion inhibitor for circulating cooling water in dynamic simulation method Y.P. Li, Y. Li & F. Liu

155

Research on the impulse current dispersal characteristics of tower grounding devices Ch.Ch. Zhu, T. Wang, X.F. Tong, P.X. Xing, Zh.Q. Feng, H.L. Lu & L. Lan

159

Innovative design of centrifugal oil press C.Q. Zhong & Y.L. Zhang

163

Analysis of the centrifugal oil press vibration C.Q. Zhong & Y.L. Zhang

167

HPLC method for determination of concentration of Nifekalant in human plasma X. Xie

171

vi

Research of over-voltage in air-core reactor using Waveform Relaxation based on Lanczos L.M. Bo, B. Bao & Y. Xu The design of automatic sorting and staking system based on PLC S7-200 L.Y. Fang, X.Y. Zhou, F.G. Wu & X. Zhang Study of dynamic optimized allocation policy for AGC regulation power based on fuzzy-Q learning algorithm H. Qian, L. Zhou, W.X. Jin & J.B. Luo Fault diagnosis of rolling bearing based on Fisher Discriminant Analysis W.B. Zhang, Ch.G. Liang & G.Ch. Li

175 181

187 195

The voltage coordinated control strategy of the power grid which includes large scale wind power R. Shi, R. Jiao, Z.J. Chi & X.N. Kang

199

Research on predict Direct Capacitor Power Control of voltage source PWM rectifier, applied to electric vehicle charging and discharging field Ch. Gong, L.F. Ma, Zh.J. Chi, W. Li, B.Q. Zhang & Y.T. Zhao

203

Development of 2500V SMB-seagull SiC JBS diodes G. Chen, Q.M. Zhang, S. Bai, A. Liu, L. Wang, R.H. Huang, D.H. Li & Y.N. Li

209

Feature-based intelligent machining method based on NX X.H. Zhan & X.D. Li

213

The application of Analyze Formability-One-Step in progressive die strip design X.D. Li & X.H. Zhan

217

The research on the signal reconstruction of the angular rate sensor L.Y. Yuan, W.G. Zhang & X.X. Liu

221

Research on the application of Beidou and GPS dual-mode timing system in the space flight tracking ship F. Zhang, Y.B. Ren & Y. Zhou

225

Study of the wind-thermal allocation ratio for wind and thermal bundled power as a source to participate in power planning X.M. Cao, T.Q. Liu, X.T. Hu, Z.H. Chen, F.J. Wang & T.Y. Guan

229

A design method of inductance-capacitance filter circuit for reducing current harmonics of high-speed motor Y.Q. Mo & P.J. Dong

235

Burr detection algorithm based on machine vision Zh. Shi, Ch.L. Xi, H.L. Li, F.Sh. Tan & J.F. Yan

243

Numerical simulation and experimental study on the anti-overload ability of cylindrical roller bearing in a short time Y.G. Ni, Y. Li, S.E. Deng & X.F. Li

249

Numerical simulation and analysis of wavefront reconstruction iterative method in radial shearing interference Y.F. Wang & Z.S. Da

255

Minimization of stator loss for high speed permanent magnet motor X.Q. Liu Iterative Adaptive Algorithm based on the cross covariance matrix of acoustic pressure and particle velocity C.R. Zhang, J.F. Cheng & B.L. Ma A method to establish a continuous operational reference station in urban districts S.B. Wang, X.J. Du, H.J. Li & W.P. Xu

vii

259

265 271

The research on the leak repairing method of the recoil mechanism N. Li, W. Jiang, H.P. Guo, S. Wang & L.M. Chen

275

Image retrieval research based on feature points and affine invariant moments B.Q. He & Z.M. Wang

279

Research on Electric Vehicle development in Beijing L. Zhang, M.Y. Pan, Z.J. Chi, Y.X. Chen & X.N. Kang

285

Lubrication performance analysis of three axial-grooved gas-lubricated journal bearing with micro grooves Y.J. Lu, F.X. Liu, Y.F. Zhang, C. Tian & M. Li

291

A new configuration of current source converter applied in HVDC J.Y. Zhao, F.M. Zhang, F.G. Liu & Y.H. Liu

297

Application of high-order grey forecast model in the short-term load forecasting X.Y. Huang & L. Yang

303

Offshore wind power scale development trends and related policy research Y. Zeng, J.S. Luo, X.L. Wang, P. Song, H.W. Huang & H.L. Bao

309

A power response characteristics equivalent model for the hybrid energy storage system P. Chen, F. Xiao, X.W. Wang, H. Yang, Z.L. Yang & L.J. Wang

313

The research for power allocation strategy of the hybrid energy storage units in distributed generation system Z.L. Yang, P. Chen, L.J. Wang, Y.H. Wang, F. Xiao, X.W. Wang & H. Yang

319

Research on the control of temperature of batch reactors by multi-media in the reactor jacket H.B. Li, D.Y. Feng & H. Liu

325

Three-phase voltage source PWM rectifier based on space-vector algorithm and one-cycle control L.F. Ma, Ch. Gong, Zh.J. Chi, J.M. Cao & L. Zhu

329

Joint positioning method with radar based on wavelet entropy H. Yu, J. Liu, M. Wang, R. Guo & Y. Yang

335

Optimal design of locking pin for surface AUVs launcher under uncertainty S.Q. Yang

341

A low temperature drift and heavy load bandgap reference voltage with adjustable output X. Xu & J. Jiang

345

Analysis of energy coupling between the computer chassis and electromagnetic pulse J. Liu, Y.F. Wang, Z.X. Chen & C.D. Yu

349

Analysis of electromagnetic shielding effectiveness of the chassis with holes under different polarization directions Q.B. Deng, C.D. Yu & Z.P. Lian

353

Analysis of characteristic parameters of pulse impact on the electromagnetic coupling of the chassis C.D. Yu, Z.X. Chen, Y.F. Wang & J. Liu

357

A new type of control method for the variable-speed wind turbines based on the PID neural network T. Li, X.Y. Hou, H.Y. Lin, Q.Y. Liu, L. Zhao & H.J. Liu

361

Research on the control method for the torque of wind generator based on data-driven T. Li, X.Y. Hou, H.Y. Lin, L. Zhao, H.J. Liu & Q.Y. Liu

365

Construction of robustly stable interval polynomial T.A. Ezangina, S.A. Gayvoronskiy & S.V. Efimov

369

A MEMS digital seismometer with new structure J. Guo & S.H. Xu

373

viii

A type of real-time vibration monitoring system based on Ethernet and RS-232 S.H. Xu, J. Guo & P.P. Li

377

Study on the transient voltage stability of distribution systems considering large-scaled dispersed EV charging Y.X. Chen, Z.J. Chi, P.W. Zheng & X.N. Lin

381

Fault diagnosis and failure rate analysis of power transformer based on cloud relation space model L.J. Guo & S.M. Tao

387

Analysis on technical innovation and regional disparities growth of China J. Xiong & W.W. Zhang

391

Author index

395

ix

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Preface

The 2014 International Conference on Mechatronics Engineering and Electrical Engineering (CMEEE2014) was held October 18–19, 2014 in Sanya, Hainan, China. CMEEE2014 provided a valuable opportunity for researchers, scholars and scientists to exchange their new ideas and application experiences face to face together, to establish business or research relations and to find global partners for future collaboration. CMEEE2014 was a most comprehensive conference focusing on Mechatronics Engineering and Electrical Engineering. The papers in this book are selected from more than 500 papers submitted to the 2014 International Conference on Mechatronics Engineering and Electrical Engineering (CMEEE2014). The book is divided into 4 sections, covering the topics of Mechatronics, Electrical Engineering, Control and Automation and Other Engineering. The conference will promote the development of Mechatronics Engineering and Electrical Engineering, strengthening international academic cooperation and communications. We would like to thank the conference chairs, organization staff, and the members of the International Technological Committees for their hard work. Thanks are also given to CRC Press/Balkema (Taylor & Francis Group). We are looking forward to seeing all of you next year at CMEEE2015. Yizhong Wang Tianjin University of Science and Technology, China

xi

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

CMEEE 2014 committee

ORGANIZER Information Science and Engineering Technology Research Association (ISET), Hong Kong, China CONFERENCE CO-CHAIRS Ai Sheng, Information Science and Engineering Technology Research Association (ISET), China Yizhong Wang, Professor, Tianjin University of Science and Technology, China COMMITTEE Jihe Zhou, Professor, Chengdu Sport University, China Hongmin Gao, Professor, Beijing Institute of Technology, China Chunguang Xu, Professor, Beijing Institute of Technology, China Haitao Li, Professor, Southwest Petroleum University, China Zhiming Liu, Professor, Liao Ning Institute of Science and Technology, China Shanglin Hou, Professor, Lanzhou University of Technology, China N.K. Sharma, Professor, The Glocal University, India Kanglin Wei, Professor, Chongqing University, China Je-Ee Ho, Professor, I-Lan University, Taiwan Chunpeng Li, Professor, Quanzhou Normal University, China

xiii

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Battlefield situation cognition system based on big electronic reconnaissance data processing S.M. Xu, H.B. Yu & F.H. Chu Electronic Engineering Institute, Hefei, China

ABSTRACT: Grid electronic reconnaissance system can be concentrated in a certain area in the layout, interconnection and interflow, share collaboration between nodes, the overall performance is high, and the battlefield environment cognitive ability is improved greatly. This paper constructs the network electronic reconnaissance system and situational cognitive system, using big data analysis and processing, to make the data more relevant and more convenient, so that the battlefield is more transparent. Keywords: 1

Electronic Reconnaissance; Big data; Battlefield situation; Cognition

INFORMATION BATTLEFIELD

2

Since the 21st century, the development of information technology has greatly promoted a revolution in military affairs, information-based war, fighting methods and operational ideas, and leading role of the military war process technology is played by information technology. Battlefield command relies on to the microelectronic technology, communication technology and computer technology as the core of the command and control system. Battlefield reconnaissance, surveillance and evaluation need all kinds of advanced electronic, optical sensing devices. Assault weapons rely more  on a variety of precision-guidance technology supported by information technology. In the information battlefield, all kinds of electronic equipments are playing an increasingly extensive role. For example, all kinds of station of radio communications equipment, loading on a variety of combat platform, always undertake the significant role for combat troops to provide command, control, communication and intelligence. Electronic reconnaissance field is an important means of battlefield situation cognition, especially to the particular area of the battlefield radio source signal to search, intercept, measure, find direction, position, monitor, analyze, recognize and get access to its technical parameters, function, type, location and purpose. Battlefield electron reconnaissance based on the real-time monitoring data and historical data fusion processing further forces the current location and trend of know each other, so as to analyze its military deployment and operational plans, etc., to achieve the enemy and know yourself. Therefore, it is of great military significance.

BATTLEFIELD SITUATION COGNITION

Situational cognitive process is a complicated human–machine system, involved in the physical domain, information domain and cognitive domain fields. The battlefield originates in the real environment of battlefield of physical domain. The collection, process and distribution of information occur in the information domain. The cognition for share situation map and decision on this basis forms in the cognitive domain, as shown in Figure 1. Physical domain is the practical condition from different fighting equipment, combat platform, command and control system which carry all kinds of electronic device, such as radio. Physical domain is the place where information comes from and decision result backs feed. Information domain contains the information perception, fusion, storage, transmission of battlefield electromagnetic radiation, and then shares the trend chart to decision maker for decision making.

Figure 1. Principal diagram of battlefield situation cognition progress.

1

Figure 2.

Data, information and knowledge hierarchy. Figure 3. Networked electronic reconnaissance system architecture.

Cognitive domain is the domain where decision makers accomplish situation perception, cognition, comprehension, inference, decision, etc. Cognitive domain exists in the decision maker’s mind and it is the source of value, belief and decision. Battlefield situation cognition accomplishes the process from collecting, analyzing and processing of the electronic reconnaissance data to battlefield situation knowledge acquisition. The meaning of data, information and knowledge is different, and also there is a connection on some level. Data is in a formalized way for a kind of fact and concept, suitable for the person or automatic device for communication, interpretation and processing, it is any meaningful or gives the meaning of the expression form, such as characters or numbers. In computing, data are input to the contents of a computer program, they through the processing of arithmetic or logical operations, obtain the result of processing. Here, compared with the word “information”, “data” refers to the source data or raw data, and “information” is defined as the data obtained through the study of the processing of data. In general, the so-called knowledge or cognition is acquired in the practice of transforming the objective world people’s basic concept, knowledge, experience and laws, and it is the foundation of human intelligence activity. From the computer science point of view, knowledge is the result of information integrated processing and, in the process of comprehensive information through mutual comparison, combined into meaningful links. As shown in Figure  2, the bottom is the data, which is the source of information and original information, express information in data can be. Information comes from processed and organized data. After summing up and summarizing, information becomes knowledge. The higher the position in the pyramid, the higher the level of abstraction. Furthermore, with the increasing level, the less the data required to express.

3

GRID ELECTRONIC RECONNAISSANCE SYSTEM

Single electronic reconnaissance equipment although is designed to be powerful, in the use process due to the frequency coverage and limited space covering function, the signal intercept probability is not high, the signal source interactivity is not sufficient, equipment between the data sharing is not strong, and correlation analysis is not enough. Therefore, the battlefield signal reconnaissance and intelligence analysis under complicated electromagnetic environment is very difficult. As shown in Figure  3, networked electronic reconnaissance system using low-cost RF-sensors, can decorate in the densely populated area, forming a grid network, and improve the traditional radio spectrum monitoring “point to surface” monitoring mode, solve the problem of the integration between the information system and realize the data interaction and information sharing. Compared with a single high-performance computer, the grid technology has a much more powerful processing capacity, the degree of information integration and sharing is relatively high. Grid reconnaissance is on the existing network technology to establish a higher level, more comprehensive resource sharing and its computing technology is more advanced. 4

BIG DATA PROCESSING

In network electronic reconnaissance system, as data are obtained continuously from all kinds of sensors, the data quantity is growing at an unprecedented rate, and due to the increasing amount of data and a variety of data format, data organization, analysis and storage have become difficult,

2

which also makes it increasingly difficult to get the valuable information quickly. At present, the data of networked electronic reconnaissance system faces several main characteristics: (1) With the full spectrum of information awareness growing demand, information awareness band continuously widened, from long to short wave, from ultra-short wave to microwave, all the way to the millimeter wave, the terahertz. At the same time, various broadband radar, the emergence of broadband communication system, also to instantaneous processing bandwidth requirement enhances unceasingly, causes the AD rate rising, thus the amount of data obtained signals intelligence reconnaissance is growing rapidly. (2) The detection signal is more and more complex, and there are many different kinds of modulation mode. According to the purpose of transmission, transmitting signals can be divided into communication signals, radar, radio fuze signal, guidance signals, navigation, etc. According to the signal, spectrum can be divided into long-wave, mediumwave and short-wave signals, ultrashort-wave signals, microwave, infrared signals, such as the laser signal. According to the way of electromagnetic wave propagation, signals can be divided into surface wave signals, ground wave, the sky wave signals, troposcatter, etc. Therefore, many complex signals make data analysis and intelligent extraction more and more difficult. (3) Huge amounts of data contain a large amount of noise or interference, as well as various signals of their own. Military intelligence reconnaissance need to find out little but useful information in the dense signal in a timely manner. (4) Military intelligence reconnaissance needs high instantaneity, ideally need real-time analytical information content, which require high processing speed. Therefore, networked electronic reconnaissance system has entered the era of big data. The author of Big Data Era, Viktor MayerSchönberger said “the real value of the big data is just like the iceberg floating in the sea, one eye can see only the tip of the iceberg, the vast majority of all hidden under the surface”. If big data is compared to an industry, the key to profit of this industry is to improve the processing capacity of the data and value-added by processing data. Also because of this, many experts say it is the era of the three pillars of data, technology and thinking. We not only need huge amounts of data, but also analyze the data of professional skills to set up innovative thinking in a unique way for the potential value of the depth of mining data. Therefore, the big data technology is applied to network electronic reconnaissance system, in order to realize the extract valuable information from large data and then for the purpose of the auxiliary decision-making. Big data technology application

in the field of networked electronic reconnaissance mainly includes the following: (1) Big data analysis is used for multi-source information fusion. It is the main task of the multi-source information fusion based on mass, multi-source, multi-type data (such as text, images, video and voice), they are related and converted to all kinds of special intelligence (e.g., Communications Intelligence (COMINT), Electronic Intelligence (ELINT), Radar Intelligence (RADINT), remote sensing information (TELINT)). (2) Big data analysis is used for relationship analysis. In grid electronic reconnaissance system, information in the process of intelligence analysis from a single node (i.e., point) is not enough to support decision making and the relationship between each unit (i.e., line and surface) is analyzed. Relationship analysis is one of the big data analysis mainly used in the field of intelligence, and eventually it generates comprehensive information support needed for the decision. (3) Big data analysis is used for situational cognition. The network electronic reconnaissance system of many sensors will produce massive monitoring data, as well as the long-term accumulation of historical data for processing and analysis. Through the combination of perception, cognition and decision support in innovative ways to use huge amounts of data, commanders greatly improve the ability to extract high value information from huge amounts of data and the cognitive ability of the battlefield environment, able to control and make decisions independently of autonomous system, fundamentally change decision model to improve the ability of rapid response. 5

CONCLUSION

A kind of grid electronic reconnaissance system is built on the concept of big data network architecture, and big data technology is used in the data processing and intelligence analysis process with the real-time monitoring data produced by many sensor and the long-term accumulation of historical data, so the cognitive ability of commanders on the battlefield environment is significantly improved through the combination of perception, cognition and decision support. REFERENCES [1] Big data in 2020[EB/OL]. [2012-12-24]. www.emc. com. 2012. [2] TERRY COSTLOW. Big Data Pose Big Challenge for Military Intelligence[Z]. Defense systems. 2012. [3] Wang Fu-rong. The analysis and application of grid radio monitoring and management, Zhejiang university of technology master’s thesis, 2013.03.

3

[4] Viktor Mayer-Schnberger. Big Data Age, Zhejiang people’s publishing house, 2013.01. [5] Yang Xiao-niu, Yang Zhi-bang. The next generation of signals intelligence reconnaissance system architecture of the concept of big data applications, Journal of China institute of electronics, 2013, 8(1): 1–7.

[6] Tang Shan-hong, Xu Hong-ru. Big data: power technology in the information age competition big data research and development new areas—the United States, Defense, 2013, 2: 73–77. [7] Wang Shan, Wang Hui-ju. Architecture: big data challenge, the present situation and prospect, Chinese Journal of Computers, 2011, 34(10): 1741–1752.

4

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Automatic recognition of waste rock in top coal caving based on digital image processing B.P. Wang & Y.J. Wang Shandon Jiaotong University, Jinan, Shandong, China

Z.C. Wang School of Mechanical Engineering, Shandong University, Jinan, Shandong, China

ABSTRACT: The recognition of the waste rock in top coal caving was investigated via digital image processing. First, the hydraulic support was processed to install the capture device. Subsequently, the images were pre-processed by median filter. Finally, the histograms were obtained and the parameters were also calculated. Three statistical characteristics which were mean, variance and energy were extracted from the histogram. The results show that the variance can be regarded as a feature to recognize the waste rock. Keywords: 1

coal-rock interface recognition; digital image procession is shown in Figure 1. Because of poor lighting at the site, the image-capture device needs to have its own lighting equipment. The device gets the images and transfers them to the signal-receiving device by the wireless transmitter. Figure  2  shows the connection diagram of the required hardware.

INTRODUCTION

Fully mechanized top coal mining is a high-yield, low-power and efficient technology. It has been popularized. In the top coal mining process, one of the problems encountered is how to monitor the degree of the caving. Now it relies on the artificial visual judgment. So security problems occur because of the dusty conditions. And also manually it is difficult to accurately monitor the top coal caving fall extent, and so sometimes the coal mining did not finish, and sometimes it mined rock. And therefore it inevitably leads to over-caving process and less-caving status. Over-caving status increases the content of waste rock, and leads to high transport costs. Over-caving will lose coal, resulting in reduced recoveries. So there is an urgent need to study the automatic identification of waste rock. Many researchers have done a lot of research on this issue, and have achieved certain results. The γ ray, infrared and radar reflection methods are used to solve the problem[1–3]. However, there is not a reliable method yet. In this paper, a digital method is discussed. 2

2.2

Image processing

The purpose of pre-processing is to make the image clear. The median filter is an effective method that can suppress isolated noise without blurring sharp edges[4]. When performing median filtering, each pixel is determined by the median value of all pixels

APPROACH

2.1 Sample image acquisition In order to obtain clean images, the hydraulic support is processed. The capture device is placed in the back of the tail beam. The location of installation

Figure 1.

5

The installation of the camera.

where N is the total pixels and ni is the number of pixels whose grey scale is ri. 3

COAL–ROCK INTERFACE RECOGNITION

The images were collected in Xinglong Zhuang Coal Mine. The median-filtered images of coal and rock are, respectively, shown in Figures  3 and 4. Contrasting the two images, some difference can be found. The coal is black, while the rock is essentially grey. The corresponding histograms are

Figure 2.

The hardware connection diagram.

in a selected neighbourhood (mask, template, window). The median value m of a population (set of pixels in a neighbourhood) is that value in which half of the population has smaller values than m, and the other half has larger values than m. Mathematically speaking, the grey-level histogram is the function of grey-level statistical properties and grey values. It expresses the proportion of areas or pixels of different grey scale in a whole image. It reflects the statistical characteristic of an image and also expresses the proportion of areas or pixels of different grey levels in a whole image[5]. Based on the significant difference in the grey level of the coal and rock, it can be obviously known that the information contained in an image of the coal or rock is very different. Three statistical characteristics,[6] which are mean, variance and energy, from the histogram are extracted as the characteristic parameters to distinguish the coal and the waste rock. Mean: µ = aver(ri)

Figure 3.

The image of coal.

Figure 4.

The image of rock.

(1)

Variance:

σ2

i

µ )2

(2)

Energy: 255

s

∑ (ri −

)3 p( p(ri ) / σ 3

(3)

i =0

A two-dimensional image is set for f(x,y), whose scales range in r0, r1, …, r255. The histogram is: P(ri) = ni /N

(4)

6

after reflection. So the variance is large. The waste rock is essentially grey. And the reflection performance is poor. So the variance is smaller than that of the image of the coal. Therefore, the variance can be as used as a feature to distinguish the waste rock from coal. ACKNOWLEDGEMENTS

Figure 5. Table 1.

Coal Rock

This work was supported by the National Natural Science Foundation of China (Grant No. 51174126), Shandong Science and Technology Development Plan (Grant No. 2013YD05005) and the Foundation Shandong Jitong University (Grant No. Z201315).

The histogram of coal.

REFERENCES

Comparison of the statistical characteristics. Mean

Variance

Energy

145.9 153.3

825.9 263.0

6.9 6.1

[1] K.W. Plessmann, B. Dickhaus and S. Scheytt 1993 Control Eng Practice. 11 457. [2] J. Asfahani and M. Borsaru 2007 Applied Radiation and Isotopes. 65 748. [3] Ren Fang, Yang Zhaojian, and Xiong Shibo 2003 Chinese Journal of Mechanical Engineering. 3 321. [4] Information on http://www.cs.ioc.ee/∼khoros2/nonlinear/median-filtering/front-page.html. [5] Xinli Song, Xifeng Zheng, Liqing Ling 2009 Chinese Journal of Liquid Crystals and Display. 24 140. [6] Jeng-Horng Chang, Kuo-Chin Fan and Yang-Lang Chang 2002 Image and Vision Computing. 20 203.

shown in Figures 5 and 6. The calculation results are shown in Table 1. It can be found that there is a larger difference in the variance than the other two parameters between the coal and the rock. 4

CONCLUSIONS

Coal is black. The grey value is relatively low in the image of coal. But part of it is bright

7

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

The application of Dijkstra algorithm based on multi-weighted distribution in GIS Shi-Min Wang, Hai-Tao Luo & Bo-Wen Wang Beijing Technology and Business University, Beijing, China

ABSTRACT: Traditional Dijkstra algorithm gets the optimal path from the source node to the destination node through iteration of path length. In fact, the shortest path cannot always meet the demand, and the path distance sometimes cannot be used as the only weight standard. This paper discusses a kind of improved algorithm of weight allocation, which redistributes the weights according to the actual macrofactors and joins in the coordinate analysis of GIS system and information such as actual traffic jams coefficient, in order to make the optimal path selection more reasonable. Keywords: 1

weight distribution; shortest path; Dijkstra algorithm 1.1

INTRODUCTION

Weight distribution

We can perform the acquisition and inputting of information in the optimization process through real-time information sharing in the GIS[4]. We classify the optimal path weights into three parts: the transportation distance, road traffic and transportation cost.

Geographic Information System (GIS) is a kind of comprehensive application system based on geography and computer science, as well as communication and space science, which is widely used in logistics, city traffic, electric power facilities repair, fire rescue etc. Its outstanding advantage is that it can plan the best route to the destination quickly according to the current geographic information, to provide reference for the rescue and driving. The path planning system is a GIS, which usually uses algorithm such as Dijkstra algorithm, Floyd algorithm and matrix algorithm[1]. The Dijkstra algorithm is commonly used in network topology, such as choosing a path in wireless mesh network[2]. Because Dijkstra algorithm adapts optimal path selection between two nodes in the network path well, it is most suitable for the application of decision in GIS system. This paper uses Dijkstra algorithm based on multi-weighted distribution in the path optimization in GIS. Most path-planning algorithms stake the shortest distance as the only standard, which are not based on the requirements of the users. With the development of economy and society, the shortest distance has been difficult to meet the requirements of the users, and people pay more attention to the optimal path problem with multifactors and multiweights[3]. Strong analysis abilities of GIS provide data to support multi-weighted path planning. After improvement, the algorithm can propose path planning scheme according to different customers’ requirements, which is more realistic and has more availability.

1. Transportation distance Transport distance directly relates to fuel consumption, vehicle wear and fatigue level of the driver, which directly determines the economic benefit. As the calculation of transport distance is simple, it is the most used index to determine the distribution routes. The time from the beginning to the end of delivery is accumulated by adding the transportation time in highway, waterway, railway, aviation, station and wharf cargo. 2. Transportation time Transportation time accounts for the majority of the whole logistic time, especially long-distance transportation. Therefore, shortening the delivery time is of decisive importance in the entire circulation time. In addition, the short ending of transportation time relies on accelerating the transportation turnover to play its full effectiveness and improve the transportation capacity. 3. Transportation cost All costs incurred in the process of carriage generally refer to the total fees that the owners cost including fees on roads, railway, aviation, transfer and other related services.

9

1.2

shortest path length from the source point to each end point, the path[n] is a collection of the corresponding path, S is the set of shortest path which has been found, U = V−S is a set of all vertices not containing the source point initially.

The path search principle of GIS system

Road traffic network GIS system in route choice, subdivided into three elements which is point, line and plane, through the combination of these three factors describes the specific physical location, size, shape and other characteristics. The point represents the converging points of two or more than two roads (cross and through). All roads abstracted into line, which is the connection between point and point. Roads and road in the line of the closed area become plane. The form of node data’s storage is point number, the abscissa, ordinate values and correlation line. The form of line data’s storage is line number, starting point, end point and the line weight. The form of surface data’s storage is plane data, the peripheral circuit data and data domain node data. Because of the above-mentioned characteristics, envisaged by the improved and combined with the spatial distribution of GIS Dijkstra algorithm, the complex traffic network path selection has a great help. 2 2.1

1. There is only a source point ‘a’ in the set S of the shortest path solved in the initialization, S = {a}. 2. Select a vertex K which has a minimum distance to a source point v from U. Add k in the set S, the selected distance is the shortest path length from the v to k, and the edge does not constitute a loop. 3. Take k as the new middle point, and modify the distance of each vertex in U. If the distance from source point v to vertex u is shorter than the original distance, modify the distance value of vertex u, so the modified distance is the addition of distance of k and the weight value from vertex k to edge u. 4. Repeat the steps of (2) and (3) until all vertices are contained in S. As the above-mentioned steps of the algorithm shows, the core step in the process of Dijkstra algorithm is selecting a minimum weight link node which is never labeled. Distribution of weights has become an important factor affecting the final result of the algorithm. In addition, Dijkstra algorithm is based on the single index (path length) to choose the path. Multi-objective path weight selection cannot be directly solved by it. It needs a corresponding weight conversion. The following Dijkstra algorithm was used to improve the single weight.

THE BASIC IDEA OF DIJKSTRA ALGORITHM The description of Dijkstra algorithm

Dijkstra algorithm is a typical algorithm based on shortest path, which is used to calculate the shortest path from a particular vertex to the other vertices in the graph or network. Its main characteristic is that it starts from the center point and expands outward layer by layer until all vertices are covered. The thought of Dijkstra algorithm is as follows: Assume G  =  (V, E) for a weighted directed graph, and divide the vertices graph in set V into two groups. The first group is the set of vertices for which shortest path has been found (it is expressed as S). There is only one source point initially, and vertex with the shortest path will be added in the set S, until all vertices are added to the S. The second group is the set of vertices for which shortest path has not been found (it is expressed as U, U = V−S, the vertices in set U are added to set S constantly until U is empty, S = V). In the process of U entering S, it must be guaranteed that the shortest path from source point to each vertex in S is less than or equal to the shortest path from source point to each vertex in U. 2.2

3 3.1

IMPROVEMENT OF THE DIJKSTRA ALGORITHM BASED ON THE DGIS The disadvantages of Dijkstra algorithm

The traditional Dijkstra algorithm simply takes the distance as the only standard of path selection. However, in real life, we have to consider the cost to achieve optimized configuration of cost and time, namely the lowest cost of logistics under the dual constraints of cost and time. 3.2

The algorithm model

The problem of multi-objective optimal path between two points of the model is shown in Figure 1. Set a directed graph, directed line segment number is s, Directed line segment representation of road links between each node, m nodes, nodes are connected by a directed edge, for each edge of directed graph, gives the multiple target weight,

The execution steps of Dijkstra algorithm

Assume n as the number of vertices in graph G  =  (V, E). The distance[n] is a collection of the

10

Figure 1.

(transportation cost, transport distance,) give a comprehensive weight for each section, and then use the comprehensive weight as the target weight, according to Dijkstra algorithm based on the shortest path, and take the shortest path as the optimal path. The second method: take a target weight of the path as the key target weight, the shortest path of N single weights can be found by the single weigh. Then calculate each target weight of the N paths. Give the weights of three sections, which are transportation distance, time and cost. Taking transportation distance as the main weight, N optimal paths will be found through the analysis of the shortest path. Then the other weights of these N optimal paths will be listed, customers can choose the path according their own will. The third method: take each target weight of the section as the main target weight in turn, and find the shortest path of the former N paths. In this paper, there are three target values in each section, so 3 N better paths can be got. Then the weights of 3  N paths are calculated, so users can choose the path which meets their own needs in the 3 N optimal paths. The first method is not only a simple algorithm, but also its shortest path is the optimal path. However, it is the comprehensive value that is assigned, because the unit of each target is different, the functional relation between the comprehensive weights and each target weight is difficult to determine. And the customer’s demands are various—different customers will have different needs in different time, so fixed coefficient is obviously not suitable. This algorithm relies more on the customer experience, and sometimes users need to adjust the relation function between the integrated weight and the target weight constantly according to the feedback information. The second method and the third method give N and 3 N optimal simple paths according to the target weight of the sections, and give the weights of each path. The algorithm is reasonable and convincing. Which is the optimal path in user’s mind is not given by the algorithm, but is selected by the user according to the calculated results. However, the second method and the third method are still different. The second method uses a target weight as the target weight to find the alternative path, and the third method treats the objective weights equally to make full use of the existing data. In the actual demand, 8–9 alternative paths have been able to meet the needs of the users. In the third method, we give three target weights, so nine paths can be given.

Multi-objective optimal path model.

they are the transportation distance, transportation time and transportation cost. The path of the selected target weight is the sum of all sections of the target weight. Out the optimal path between any two nodes in the graph in accordance with customer demand. 3.3

Improvement of the algorithm

According to the analysis of the multi-weights, we can use three methods to improve the algorithm. For the three factors of path (i, j), we record transport distance as L, the transportation time as T and integrated transportation costs as S. For the path, assume that the evaluated value of path (i, j) is P, and we can get: Pij = m1L + m2T + m3S as the scale of addition, in which m1, m2, m3 are constants and can be adjusted according to different constraints. There is a path between any two nodes i and j including n sections, so the weights of Pij can be expressed as k

Pij = m1 ∑ n =1 L + m

k

∑ n =1 T

k

m3 ∑ n =1S

In GIS, collection of edge information can be represented by the edges of road length, average speed and vehicle fuel consumption, and the edge of the weight is made up of these information. A path distance, driving along this path of oil consumption and so on are all terms by super imposing the corresponding weights are calculated. For along this path the time required, can first use the average speed of the edges of each edge length divided by, then find the required on each side. When asked, finally put these time superposition. That is to say in the GIS, most constraint information have the additivity. The first method: simplify some weights of the section. Considering various factors of the section

3.4

The advantage of improved algorithm

There is not only one factor of length in the problem of path planning, but there are still many

11

Table 1.

Nine alternate sections weight of the path.

Main target value

Source point

Middle vertex

End point

Mean of transportation distance (meters)

Transportation distance

1 1 1 1 1 1 1 1 1

2-7-18-24-33-36 2-7-17-24-29-36 2-7-18-24-30-35 2-7-18-24-33-36 2-7-18-24-30-35 2-5-14-21-25-31 2-7-17-24-29-36 2-7-18-24-33-36 2-7-18-24-30-35

37 37 37 37 37 37 37 37 37

1324 1359 1367 1359 1367 1372 1359 1324 1367

Transportation time Transportation cost

factors that need to be considered. Some users pay attention to transportation time, while some users focus on transportation costs, and usually other factors such as roads also need to be considered. After the algorithm improvement based on multiweights, the path selection is more reasonable. To classify the weights according to the actual macroeconomic factors can meet the needs of different users. 4

got 9 alternative paths initially, there are few alternative paths eventually. In order to increase the number of final alternative paths and make user have more non-quantitative factors in the final decision, when the final alternative paths are few in number, we can increase the number of k under the condition of constant m, in order to make the number of final output conform to the needs of the users. This paper analyzes the current situation of optimal path research, based on the previous research, and it obtained the following results: 1) This paper analyzes the models of optimal path and related algorithm combined with the actual situation of our distribution center, and also puts forward the multi-weight optimal algorithm. 2) In the multi-weight optimal algorithm, there are not only fixed target weight but also random variable target weight in each section (fixed mean and variance, but not necessarily limited to normal distribution), in order to expand the optimization goal further and more accord with the reality. 3) It puts forward the method to solve the problems existing in multi-weight optimal algorithm. Especially for the algorithm in which each target weight plays the role of weight in turns, this algorithm can make full use of the data to provide multiple paths, and the algorithm is less complex compared with the algorithm based on single target when providing the path of the same number. The paper lays a good foundation for further study on integrate algorithm of the GIS.

REALIZATION AND RESULTS OF THE MULTI-WEIGHTED OPTIMIZATION ALGORITHM

We searched road images of a city in China via Google earth, and made a statistics on the road information according to GIS. We get 37 nodes, 117 sections, in which the starting point is 1 and the end point is 37. We processed each target weights. After processing, we can get transportation length, transport costs, and transport time mean and variance. We took the path length and the cost of the vehicle through as a fixed value, and the time of the vehicle through a section as a random normal variable. We took path length, transportation cost and transportation time as target values in turns to obtain three simple shortest paths. This paper assumes that transportation time is not relevant, so it is convenient to estimate the variance of transportation time after getting the nine paths. We take the third method as the basic algorithm, run Dijkstra algorithm mainly based on transportation length, and get nine alternate sections correspondingly. The point and the target weight of the path are shown in Table 1. The results of the third method not only gives nine paths, but also lists the transportation distance, transportation cost, and the mean and variance of transport time for each path. Due to there are repeated path in seeking 3 simple shortest path switch different target weight, though we

REFERENCES [1] Gu ling-lan, The Optimization of Dijkstra in GIS Route Analysis. Department of computer Engineering, 2006, 34(12):54–56. [2] Tang wen-wu, Shi xiao-dong. The Calculation of the Shortest Path Using Modified Dijkstra Algorithm in GIS. Journal of Image and Graphics. 2000, 5(12):1020–1023.

12

[3] Shang jing. On the Shortest Path in Direction Diagram byDud Weight. Journal of Institute of Financial and Commercial Management. 2008, 10(2): 21–22. [4] Zhang mei-yu, Jiancheng-feng. Research on Dijikstra algorithm in optimal path with multiple constraints of agricultural product distribution. Journal of Zhejiang university of technology. 2012, 40(3): 322–330. [5] Fan yue-zhen, Jiang fa-chao, Design of Vehicle Optimization route Algorithm. Department of Computer Engineering. 2007, 28(23): 5758–5761. [6] Zhang xin-yi. Wu jin-pei. An Implementation of Path Planning Algorithm Applied to Vehicle Location and Navigation System. Computer Automated Measurement & Control. 2001, 9(4): 16–17.

[7] Li yuan-chen. Liu wei-qun. Analysis of the Shortest Route in Network on Dijkstra Algorithm. Microcomputer Applications. 2004, 25(3): 296–298. [8] Hao wei. Liu wan-qing. Shortest path algorithm for rescue Vehicles based on GIS. Computer Applications. 2008, 28: 104–108.

13

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Design of battery charging and discharging circuit H.Q. Zhan & Z.P. Li Jiangxi Science and Technology, Normal University, Nanchang, Jiangxi, China

ABSTRACT: This study reports on the design of charging and discharging circuit based on operational amplifier. Battery management is divided into charging (constant current and constant voltage), discharging and protection (overvoltage, under-voltage, over-current and over-temperature). At initial charging, the battery uses constant current charging mode. When it reaches the rated voltage, it will switch to the constant voltage charging status automatically. The battery supplies electrical power. When the discharge voltage is less than the minimum voltage, it will convert to under-voltage protection state. If discharging is stopped, then it converts to the charging state. It can convert to the state of charging by artificial conversion too. The design achieves the requirements of the subject well. The circuit design is brief and clear and has a high performance price ratio. Keywords: battery; control circuit of charge and discharge; operational amplifier circuit; circuit protection 1 1.1

SCHEME DEMONSTRATION AND ANALYSIS

current closed-loop control is shown in Figure  1. According to the concept of virtual short of integrated op-amps, we can get it available:

The choice of battery charging scheme IL = Vi/R1.

Scheme one: Using a 555 timer to get a monostable triggered PWM circuit to control the duty ratio of PWM wave. Thereby it can achieve the control of main circuit of charging on–off time and change the average charging current. This scheme uses the battery voltage as the power supply voltage of 555 directly. The voltage fluctuation is large, which is easily go against 555 timers work. And the design of equalizing charge and floating charge is complex. Scheme two: This battery charging circuit design is based on the op-amps and triode. It uses the LM317 1.25 V output, and the end pressure drop to achieve the purpose of filling and floating. Compared with reference voltage, battery voltage output of the high and low level controls the triode conduction, which switches the filling and the floating. This scheme of voltage has a certain error with the theory voltage. Scheme one can better meet the system requirements, but the control circuit is complex. Although the second scheme design parameters have a certain error, the impact is very little on the system. So we chose scheme two. 1.2

where IL is the load current, R1 is the sampling resistor and Vi is the input signal of op-amps. If R1 is fixed, IL is completely determined by Vi. At this time no matter how the Vcc or RL changes, we can make IL to remain stable by using automatic adjustment function of feedback loop.

The choice of battery discharge scheme

Scheme one: We use constant current closedloop to control the circuit. The typical circuit of

Figure 1.

15

Constant current control.

Scheme two: Adopting the method of constant voltage discharge, compared with the reference voltage, makes the output voltage constant. But for battery, adopting the method of constant current discharge can improve the efficiency of the battery. The greater the current of battery is, the lower the efficiency of the battery is. In view of the above-mentioned analysis, this system adopts scheme one. 2 2.1

SYSTEM THEORY ANALYSIS AND CALCULATION

Figure 2.

System block diagram.

Constant current, constant voltage charging circuit parameter design

As it is shown in Figure 3, during the early charging, the battery is charged with 0.25  A constant current. According to the three terminal voltage regulator tubes which have 1.25 V of voltage drop between output terminal and earthling terminal, we can work out the following equation: R2 = 1.25/0.25 = 5 Ω. When the voltage of battery reached 7 V by constant current charging, it switches to constant voltage charging. Op-amp reference voltage is 3V. When the partissal pressure value of the R8 and R11 is greater than 3V, the charging of battery is converted into a constant voltage. If we set R8  =  10k, then R11 is a value for 10k potentiometer. 2.2

Figure 3. Constant current, constant voltage charging circuit.

The parameter design of commercial power, solar power charging switching circuit

As it is shown in Figure  4, R2  =  20k, R4  =  20k, partial pressure value of R4 is 0.5VC2 (VC2 is battery voltage), as a reverse input of the comparator. According to the characteristics of photosensitive resistance, the partial pressure of resistance R5 is VREF. As a positive input of the comparator, it switches to the solar charge during the daytime and adopts commercial power charge in dark environment. 3

THE SYSTEM CIRCUIT ANALYSIS

Figure 4.

According to the requirements and scheme demonstration, the block diagram of this system is shown in Figure 2. System working principle: At the early charging, the battery voltage is less than 7.0 V, the signal through the logical processors output low level. That means the base of Q1 and Q2 for the low level and makes the Q1 is on conduction,

Charging mode switching circuit.

Q2 is cut off. At this point, the charging process is at the constant current charging stage. When charging voltage is higher than 7.0 V, then Q2 is on conduction, and charging process converts to constant voltage charging. Adjusting R11 can change the battery voltage during constant current charging. R3 is the resistance of negative

16

temperature coefficient, which functions on temperature compensation so as to improve the accuracy of constant-voltage charging voltage. When constant voltage of charging reach to the rated voltage, the power supply to the load condition, which can also be manually switched. This system can achieve the purpose of solar charge during daytime and commercial power charge in dark environment. 3.1

Constant current, constant voltage charging circuit

The LM317 is three-terminal integrated adjustable output voltage stabilization block, with over-current protection and thermal overload protection. When battery voltage is less than 7.0 V, op-amp U1 and U2 output low level, Q1 is on conduction, Q2 is on deadline. At this time, it is at constant current charging stage. When battery voltage rises to 7.0 V, then the charging process is converted into a constant voltage charging. And adjustable R11 can change the charging voltage. R3 has the function of temperature compensation, which can improve the accuracy of constant voltage charging. And the circuit is shown in Figure 3. 3.2

Under-voltage protection circuit.

Figure 6.

Battery discharge circuit.

Charging mode switching circuit

This system design for battery charging methods are solar charging and commercial charging. They can be both automatic and manual switched. This switch circuit is shown in Figure 4. When there is light irradiation, VREF reference voltage is less than 0.5VC2 voltage, signal through logical processors output high level, then relay normally open switch is closed and the normally closed switch is open. The battery switches into solar battery charging. If in the dark environment, relays also switch to the original state which is commercial power supply. This circuit can also be manually switched. When close button S1, relay normally open switch is closed, it’s on solar power charging, When close button S2, it switches into a commercial power charging status. 3.3

Figure 5.

3.4

Battery discharge circuit

This system design of discharge circuit adopts the way of constant current circuit. This circuit is shown in Figure  6. The load current through the resistor R9 obtained by U2 amplification 4 times sampling voltage feedback to the amplifier U1, compared with 3v reference voltage. The load constant current value is 1.5 A. By changing the reference voltage value to change the constant current value.

Battery under-voltage protection circuit

Battery adopts the constant current method supply current to the load. As the battery discharges, the battery voltage is slowly reduced. When the battery voltage is less than 5 V, signal through logic processing and output high level, then relay normally open switch is closed. And the battery is no longer discharge but switches into charging status to make the battery storage voltage. Under-voltage protection circuit is shown in Figure 5.

4

CONCLUSION

We adopt the charging way by using the change of the battery voltage to control the charging

17

circuit directly. It can realize two charging ways: solar power charging and commercial power charging. In the initial stage of storage, battery adopts constant current charging. When it reaches rated voltage, it adopts constant voltage charging. And when full of voltage, the battery can provide the load voltage with the method of constant current. This system completed the topic request, not only can charge and discharge, but also can undertake solar charging. The design of this circuit is brief and has a high cost performance.

REFERENCES [1] Bai, S.T. & Ying, C.H. 2001. Analog electronic technology foundation. Beijing: Higher Education Press. [2] Qun, F.Z. & Song, Q..L & Jin. S. 2004. A battery voltage equilibrium system. Electrical automation 04(4): 94–99. [3] Feng, G.J. & Ming, H.W. 2008. Long life electric bicycle with VRLA battery research. Battery 08(7): 14–19. [4] Bin, L.L. & Peng, C. & Lin, H. 2008. Protection and restoration of lead-acid battery technology research. Battery 08(1): 25–28.

ACKNOWLEDGEMENT Thanks to Zeping Li for the corresponding author of this paper.

18

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Automatic flight control of a certain UAV using LQG design SH. Y. Zhang Department of Aerial Instrument and Electrical Engineering, The First Aeronautical Institute of Air Force Xin Yang, China

ABSTRACT: This paper presents the design and simulation of automatic flight controllers for a UAV by using a Linear-Quadratic-Gaussian (LQG) control approach. The automatic flight controllers are constructed based on two independent LQG regulators which govern the longitudinal and lateral dynamics of the UAV respectively. The LQG design is able to retain the guaranteed closed-loop stability while having incomplete state measurement. The Kalman filter is used to provide the estimated states. The flight-test results show that it is highly feasible and effective with a relatively simple onboard system. Keywords: (LQG) Linear-Quadratic-Gaussian; UAV (Unmanned Air Vehicle); automatic flight controller 1

2

INTRODUCTION

THE UAV MODELING

The UAV is considered as a single rigid-body, for a certain UAV system, the following are defined for the longitudinal model:

In recent years, UAV technology has experienced a rapid growth in popularity. As more countries around the world begin to acknowledge and actively pursue UAV systems to augment their own Intelligence, Surveillance, Target Acquisition and Reconnaissance (ISTAR) capabilities, UAV has been one of the latest military and force-multiplying technologies pursued by many countries around the world. Fixed-wing UAVs, especially the larger ones with gross weight in excess of 100  kg, own distinct advantages over other configurations in terms of payload capacity, operational speed, altitude, range and endurance. This paper focuses on the automatic control strategies developed exclusive of conventional fixed-wing UAV systems. The linear-quadratic regulator, omitted as LQR (1) has following advantages (1) control theory is well-established mathematically and the resulting control law is simple and elegant; (2) the optimal control gains are automatically generated in the solution of the control equations; (3) highly suitable for the automatic flight control of UAVs like the stability and control of a high-altitude, longendurance UAV. However, the biggest pitfall of LQR is the requirement of the real-time full-state measurement which is often unavailable in practice (2). The Linear-Quadratic-Gaussian (LQG) controller is an extension of the LQR where the unmeasured states are estimated using an optimal observer, i.e. the Kalman filter. This gives the LQG the advantage of dealing with the uncertain linear systems disturbed by the additive white Gaussian noise while having incomplete system state information available for the control-loop feedback.

x [ ΔV , Δα , Δqq , h ]T T y [ ΔV , Δq h] u = [ Δδ e ]

(1)

where V, α, q, θ, δe are the airspeed, angle-ofattack, pitch rate, pitch angle, altitude and elevator deflection respectively. Similarly, for the lateral model, x [ Δβ Δppp, Δr, r, Δφ ]T y [ Δp, Δr, Δφ ]T u = [ Δδ a , Δδ r ]

(2)

where β, p, r, φ, δa, δr are the sideslip angle, roll rate, yaw rate, bank angle, aileron deflection and ruder deflection respectively.

3

LQG REGULATOR DESIGN(3)

LQG Regulator is a combination of the LinearQuadratic Regulator (LQR) and the Kalman filter. u(K) = −Kx(K)

(3)

where the optimal constant feedback gain, K is obtained such that the quadratic cost function J is minimized.

19

Figure 1.

∞

J

LQG regulator.

(k ) ∑ x(k

T

Qx( k k)) + u( k )T Ru( k )

(4)

k =1

where K is readily attainable by solving the socalled discrete algebraic Ricatti equation (4) for P: P Q + AT PA ( AT PB )(R + BT PB ) 1( BT PA) (5) which yields K

(R + BT PB ) 1( BT PA)

The block diagram of LQR regulator is shown in Figure 1. 4 4.1

FLIGHT TEST SIMULATION

Figure  2. Output response of straight-and-level automatic flight.

Straight-and-level (equilibrium) flight

Keep throttle level unchanged throughout all automatic flight modes, Straight-and-level automatic flight are carried out in a gradual, successive schedule. It is evident that the closed-loop control systems for both longitudinal and lateral dynamics are stable during the entire automatic flight lasted for 27.5 seconds. The variation of the altitude is bounded between 276  m and 282  m which means the system error is generally less than 3 m. Moreover, despite having only one control input (elevator deflection) in the longitudinal controller, the airspeed is able to be maintained within an error of 4 km/h from the trim airspeed, 104  km/h. No distinct oscillation was observed on both airspeed and altitude responses as well. On the other hand, after a short transient process, the bank angle response enters steady state with bounded error generally confined within +5 deg. This indicates that the controller is able to maintain the wing at level position under external disturbance (wind gust and aerodynamic load). In addition, the equilibrium condition of the aircraft in flight is being sustained using little

Figure 3.

20

Simulation of bank angle tracking test.

control effort. The figure shows that the elevator and aileron typically operate in the order of 4 deg and 2 deg deflections respectively while the rudder is practically unused. This is one of the strength of linear-quadratic control approach where the compromise between control effectiveness and control effort is accomplished with the proper selection of Q and R weighting matrices. 4.2

separately designed. Flight test simulation have shown that the controllers perform very well even though simplified linear models are used in the synthesis of the controllers. In addition, the controllers exhibit exceptional stability and tracking performance with good disturbance rejection capability. No evidence of significant cross-coupling dynamics between longitudinal and lateral motions was found. Also, the resulting control laws are simple and efficient enough to be easily realized using limited onboard computing power.

Bank angle tracking

The desired square wave has an amplitude and period of 20 deg and 10 sec respectively. As shown in Figure 3, the bank angle response exhibits excellent transient and steady-state properties. Typically, the overshoot percentage, rise time and settling time lie in the order of 20%, 1.5 s and 3 s respectively. The damping of system is more than satisfactory where no noticeable oscillation occurs before the tracking error converges to zero. 5

REFERENCES [1] Hsiao, F.B. Engine speed and velocity controller development for small unmanned aerial vehicle, J Aircr, 2008, 45(2), pp 55–65. [2] Lee, C.S., Hsiao, F.B., Jan, S.S. Design and implementation of linear-quadratic-Gaussian stability augmentation autopilot for unmanned air vehicle, Aeronaut J, May 2009, 113(1143), pp 275–290. [3] Li Li, Zhang Hua-min. Neural Network Based Feedback Linearization Control of an UAV, AEIT2011 Conference, pp 45–48. [4] Ogata, K. Discrete-time Control Systems, 1987, Pretice-Hall. [5] Erdos, D., Watkins, S.E. UAV autopilot integration and testing, 2008, IEEE Region 5 Conference, pp 1–6.

CONCLUSIONS

This paper presented a design and simulation of automatic flight controllers based on the LQG theory on a fixed-wing UAV. The automatic flight is achieved by utilizing independent linear longitudinal and lateral controllers which are

21

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Research on online and remote calibration technology of electronic belt scale based on weights superposition method X.P. Shang, H.Y. Chen & Z.W. Huang Zhejiang Province Institute of Metrology, Hangzhou, China

ABSTRACT: On the basis of research in existing measuring instruments and measurement method of coal, we explored online and remote calibration technology of the main measuring instruments— electronic belt scale. This system is based on incremental superposition method. It uses weights which simulate actual material to superimpose on the auxiliary belt scale’s weighing sensors, as theoretical superposition. By comparing the cumulative difference between the main and auxiliary belt to the theoretical superposition, we can obtain calibration results. The entire calibration process can be completed online and all real-time data can be transmitted to the high compatibility and high reliability intelligent measurement network platform remotely. The system provides the technical means of online and distance calibration of belt scale for measuring departments and enterprises and it saves a lot of manpower and resources and improves the efficiency of calibration and ensures that the belt scale works accurately. Keywords: 1

electronic belt scale; remote calibration; weights superposition

INTRODUCTION

regulation’s requirements, in order to determine the specifications of belt scale in conformity with the metrological requirements or not, calibration of belt scale is needed. Calibration methods of belt scales are divided into physical calibration methods and simulation calibration methods. Physical calibration methods include hopper scale calibration method and material superimposed calibration method. Simulation calibration methods include cyclic chain method, roller chain calibration method, weights superposition calibration method and so on. This article presents a remote calibration system of electronic belt scale based on weights superposition calibration method. This system can reduce costs of calibration of belt scales in use for enterprises and reduce economic losses from trade settlement for companies which have coal as main energy. For energy management department and statutory metrology department, this remote calibration system can realize the real-time monitoring and forecasting of energy data.

In our country, coal is the main energy. According to statistics, China’s primary energy consumption has been more than 4 billion tons of coal equivalents, of which 3.5 billion tons are coal that accounts for 74.7% of the energy consumption. Modes of coal transportation include railways, shipping and automobile transportation, and shipping becomes the main way for its lowest costs relatively[1]. Coal which is transported by ship is measured mainly by water gauges and belt scales. Water gauge measurement is affected more by waves, temperature and the observer’s capability, and its repeatability and reliability is bad. Furthermore, water gauge measurement is unable to trace. Belt scales are stable, reliable and traceable in measurement and its maximum permissible error can be controlled in the 0.5% range. Belt scales are also widely used in enterprises which use coal as the main energy source. In addition, the electronic belt scale measurement can realize on-line measurement of coal and improve work efficiency and it has achieved more and more attention[2]. As a kind of measuring equipment which weighs continuously and automatically when belt conveyer is conveying solid bulk materials, electronic belt scale’s accuracy is not only associated with the product itself, installation position and installation quality, but also closely linked to periodical calibration[3]. According to the national verification

2

OVERVIEW OF THE DEVICE

The calibration method based on weights superposition of belt scales belongs to a kind of simulation calibration methods. The device includes belt conveyors, the major belt scale, the auxiliary belt scale, weights superposition structure, IPC and so on. Figure 1 is the structure of the belt scale.

23

information technology. Electronic belt scale based on incremental superposition method has the advantages of high efficiency for calibration and the calibration process being automatic and online which are very important for realizing belt scale’s remote control for calibration. The calibration system based on weights superposition of belt scales makes use of the Internet and other technical means to develop a Coal Measurement Network Platform which can realize online monitoring and remote calibration. Remote calibration methods of the system are realized by the way that the web server communicates with a local webcam and the calibration device’s IPC in the same time. The local IPC has installed the calibration software which is designed and developed by us. This software can send commands directly to the PLC to make parameter setting, to drive motor, to collect and transmit data. Meanwhile, from the database founded by the Coal Measurement Network Platform, we can inquiry the history of calibration results of the local computer. Figure 2 is the data flow chart of the remote calibration system. The Coal Measurement Network Platform is a part of “Energy Smart Metering Data Service Platform” and its main functions are measurement apparatus management, remote calibration control, monitoring online and statistical diagnosis and so on. First, we make surveillance cameras and IPCs which control PLCs to access the Coal Measurement Network Platform. The Coal Measurement

Figure 1. The chart of the belt scale based on weights superposition.

The basic principle of the calibration method based on weights superposition of belt scales is as follows: we use weights which simulate actual material to superimpose on the auxiliary belt scale’s weighing sensors as theoretical superposition when coal or other solid bulk materials are being conveyed. By comparing the cumulative difference between the main and the auxiliary belt scales to the theoretical superposition, we can get the calibration results. Before you start weights superposition calibration, you must first ensure that the main belt scale’s and the auxiliary belt scale’s error conditions and accuracy are consistent, namely scale– scale comparison experiments are needed. The principle of the scale–scale comparison experiment is as follows: in the main and the auxiliary belt scales weighing the same amount of material conditions, we amend measurement coefficient of the auxiliary belt scale while the main scale is the standard to insure that the main and the auxiliary belt scales get the same material cumulative values. When a calibration is started, we need start weights superposition structure to put standard weights on the auxiliary belt scale’s weighing sensors. After a period of time, we calculate cumulant difference between the main and the auxiliary belt scale within the set time and then compare the difference to the theoretical superposition to obtain calibration results. At the end of the calibration, we start the weights superposition structure again to lift the standard weights. 3

THE REMOTE CALIBRATION SYSTEM

Remote calibration technology is the inevitable result for combining of the information technology and traditional measurement and testing technology. Remote calibration is for distance calibration in the use of information networks and communication networks which is a multi-disciplinary comprehensive cross involved in instrumentation, metrology, computer hardware and software technology, and

Figure  2. The data flow chart of the remote calibration system of belt scales based on weights superposition methods.

24

have developed Windows service and web service separately. Windows service gets commands stored in the database periodically, and then sends commands through the web service to related metering monitoring equipment. During the calibration process, we need to control the hopper’s opening or closing, and the belt conveyor’s starting or stopping, and weights superposition structure’s lifting and falling, as shown in Figure 3. The Coal Measurement Network Platform sends control command to the web server by the Windows service. After the web service in the web server receives the commands, it forwards the commands to the IPCs, the cameras and other equipments depending on the control objects. PLCs receive commands stored in the IPCs’ web service, and then control all components’ motors to move. By adjusting the monitors’ angle, operators can monitor metering equipments’ operation status remotely and in real time. Meanwhile, cumulative flow of belt scales and speed of conveyors and other online metering data will send to the Coal Measurement Network Platform remotely. 4

CONCLUSION

This paper has successfully developed remote calibration system of electronic belt scales based on the weights superposition method. It provides technology methods of belt scales’ remote and online calibration for measuring departments and enterprises, saves a lot of manpower and material resources, improves the calibration efficiency and ensures the accuracy and reliability of belt scales’ online measurement. ACKNOWLEDGMENTS This work was financially supported by the major science and technology project of Zhejiang province (2012C01026-1).

Figure 3. The control process of belt scales’ calibration based on weights superposition methods.

Network Platform realize belt scales’ on-line monitoring and remote calibration by controlling cameras, video recorder hard drives, and IPCs. Remote surveillance uses dot-IR cameras which has scanning frequency 50  Hz. In the Coal Measurement Network Platform, the web system can send control commands to PLCs and store these commands in web database. Basic calibration information of belt scales, including models, manufacturers, precision and flow parameters, is stored in database. Web servers and PC machines (or IPCs)

REFERENCES [1] Simpson J. 2000. Canadian Weights and Measures and 0.1% Certified Belt Scales. Canadian Weights and Measures (Technical Paper). [2] Wang Z.J. 2009. The calculation method of hanging code calibration of electronic belt scale. Journal of weighing apparatus 38(7): 36–37. [3] Chen Y.F. 2010. The development of dynamic chain code loop checking device of electronic belt scale. In Shanghai, east China university of science and technology.

25

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

The application of wavelet filter in static synchronous compensator Li Zhang School of Electronics and Information Engineering, Anhui University, Hefei, China

Ning Liu Heifei Normal University, Hefei, China

ABSTRACT: Reactive power balance of power system is important. The static synchronous compensator (STATCOM) based on inverter technology is a new dynamic reactive power compensation, and has many advantages compared with Static VAR Compensator (SVC), so STATCOM has vast market prospect. In this article, wavelet filter is applied in STATCOM to filter the harmonics of load current. The active portion and reactive portion of the fundamental current are calculated by dq transformer, and the reference signal of output current is calculated by dq reverse transformer, using the reactive portion and the output of PI controller which controls the DC voltage, and PWM tracking control technology is employed. The simulation results indicate that the application effect is good. Keywords: 1

wavelet filter; STATCOM; reactive power compensation; simulation in 1998  made by Siemens, Germany. In China, the STATCOM device of ±20 Mvar was run in Chaoyan transformer substation of Henan power company in 1999, and in 2005 the STATCOM device of ±50 Mvar was installed in Huangdu Xijiao transformer substation, which was used to improve the voltage stability of 220 kV bus and inhibit the bus voltage surge.[2–3] In this article, the wavelet filtering is applied in STATCOM, which filters the harmonics in voltage and current and get the fundamental voltage and current. Simultaneously, the active portion and reactive portion of the fundamental current are calculated by dq transformer, and the tracking control technology is employed to control the output current of STATCOM, using reactive portion of the fundamental current as reference signal. The output current tracks the reference signal and the source current does not contain fundamental reactive current.

INTRODUCTION

Reactive power in power grid is the basic reason of the deviation of power system voltage from the nominal value. When the voltage deviation gets big, the electrical equipment performance would be deteriorated, which may lead to not only the low operating efficiency of equipment, but also the damage caused by the overvoltage or overcurrent. Big voltage deviation is the thread to the power system’s stability and influences the system’s economical operation. The necessary and sufficient way to ensure every node voltage of power system in normal levels is that the power system has plenty of reactive power sources and essential voltage regulation method is used. The SVC and STATCOM have dynamic reactive compensation properties in many kinds of reactive power sources. Compared with SVC, the STATCOM has many superior performances, such as fast governing speed, wide running range, and can greatly reduce the harmonics in the compensation current when the PWM control technique and so on are used. Besides, both the bulk and the weight of the device are decreased because the reactor and capacitor used in STATCOM are far smaller than these in SVC. Japan and the USA, respectively, developed a set of 80 Mvar and 100 Mvar STATCOM device adopted GTO thyristor in 1991 and 1994. Both the devices were successfully put into commercial operation.[1] The STATCOM device, with a unit capacity of 8 Mvar, was also put into operation

2

THE STATCOM STRUCTURE AND WORK PRINCIPLE

The schematic diagram of STATCOM adopted voltage source inverting circuit is shown in Figure 1. The voltage source inverting circuit gets into the power grid through the transformer, which can match the source voltage with the grid voltage. If there is no need to match the two voltages, the inverting circuit can also get into the grid through the inductance. In order to

27

Figure 1. The STATCOM with voltage converter circuit.

achieve dynamic reactive power compensation, the output voltage phase and amplitude of the AC-side of convert circuit can be controlled (or directly control the current of AC-side) to make the fundamental reactive current ((iiac), which is in the converter circuit, equal with the load fundamental reactive current (ial), and there is no fundamental reactive current in (i the source current (i (ias ). More thyristors used in series can improve the STATCOM capacity and the voltage grade, which can also be achieved through the multiple technology and multi-level technology. 3

Figure 2. Table 1.

THE THEORY OF WAVELET FILTERING

The formula of f(t)∈L2(R) is launched following the space combination shown in equation (1),[4–5]

The Frequency band and harmonics.

Wavelet

Frequency band (Hz)

Harmonics

cd1 cd2 cd3 cd4 cd5

1600–3200 800–1600 400–800 200–400 100–200

32–64 16–32 8–16 4–8 2–4

J

∑ Wj

L2 (R ) =

VJ ,

(1)

tains base wave after the five layers of wavelet decomposition.

j =−∞

where J is an arbitrary scale. So, the f(t) can be written as f t) =

The wavelet decomposition.

J

∞

∑ ∑

j =−∞ k =−∞

dj k

j k (tt))

+

∞

∑

c j kφ j k (t ).

4

CONTROL OF STATCOM

Block diagram of fundamental reactive load current’s detection and power grid reference current’s generation is shown in Figure 3, where sine and cosine signal (sinωt, cosωt) are obtained, which have the same frequency and power with the power grid by pll. ial , ibl and icl are the load current, and the fundamental current (iiaf , ibf , icf ) can be get after wavelet filtering described in step 2. The three-phase fundamental current is assumed as follows:

(2)

k = −∞

The signal f(t) is decomposed by five-layer orthogonal wavelet (Fig. 2). If the signal fundamental frequency is 50  Hz, according to Shannon’s sampling theorem, the fundamental signal, sampled 128 points per cycle, can be analyzed to 64 harmonics (3200 Hz). If the maximum frequency of f(t) is 3200  Hz, according to Figure  2, the band division on the wavelet decomposition is shown in Table  1,[1] where each decomposition of the signal or low frequency (ca) of the band is down into some low-frequency band (ca) and high frequency band portion (cd), and the decomposition of the band is in accordance with the highest frequency that can be analyzed. The 128-point sampling is according to the band decomposition at the maximum frequency of 3200 Hz. In Table 1, the voltage signal and current signal are for five-layer wavelet decomposition. If the voltage and current without DC and second harmonic, their low frequency (ca5) only con-

iaf

I m sin(ωt − ϕ )),

(3)

ibf

I m sin(ωt − 2π / − ϕ ),

(4)

icf

I m sin(ωt + 2π / − ϕ ).

(5)

The dq transformer of three-phase current is shown as follows,[1,6] so that the load fundamental current active component and reactive component can be obtained. In order to maintain the DC bus voltage (UC) stable, closed-loop control of capacitor voltage is used, whose controller is the PI regulator. I′P is assumed as the output of PI regulator,

28

5

The power supply system with STATCOM is shown in Figure  5, where the three-phase power is symmetrical. The STATCOM and its control circuit are at the right part of the figure. At the left part, there are access point detection of voltage and current, the three-phase power and its load which is the three-phase full-controlled bridge rectifier circuit. The three-phase power supply is joined by the starconnecting and the equivalent inductance of the power is LS, while the equivalent resistance is not drawn. The current whose characteristic harmonic is 6k ± 1 (k = 1, 2, 3, …), causing the supply point voltage distortion, is formed by the three-phase fullcontrolled bridge rectifier circuit. The STATCOM is the three-phase bridge voltage inverter circuit. It is connected to the access point through the inductor LC, whose effect is filtering. The three-phase voltage inverter circuit is controlled using the hysteresis current tracking control technology. The simulation system is to maintain DC voltage (voltage across the capacitor) of STATCOM stable and compensate the load fundamental reactive current. The grid voltage and load current contain harmonics, which can be filtered by wavelet transform. The signal is decomposed by five-layer orthogonal wavelet, and the low frequency part (ca5) is reconstructed. According to the frequency division of wavelet decomposition in Table 1, the lowest harmonic frequency in the signal is 5, and the reconstruction signal of the low frequency part is just the fundamental signal. So, the lowest harmonic of the voltage and current in the simulation system is 5 and both the voltage and current are decomposed by five-layer wavelet, and the fundamental voltage and current can be obtained by the low frequency part’s reconstruction. The simulation results are shown in Figures 6–9. In the four figures, the control angle of the full-controlled rectifier bridge has been changed from 5° to 60° at 0.4 s. Figure 6 shows the three-phase voltage (uap, ubp, ucp) of the public power supply point (PCC) and its wavelet filtered voltage (ual, ubl, ucl). Figure 7 shows

Figure  3. Block diagram of the power grid reference current’s generation.

Figure 4.

Tracking control.

⎡ 2 ⎞ 2 ⎞⎤ ⎛ ⎛ sin ωt sin ωt i ωt + π ⎥ π ⎡ IP ⎤ 2 ⎢ ⎝ ⎠ ⎝ 3 3 ⎠⎥ ⎢ ⎥= ⎢ 2 ⎞⎥ ⎛ ⎢⎣ IQ ⎥⎦ 3 ⎢cos ωt cos ⎛ ωt 2 π ⎞ ωt + π ⎥ ⎢ ⎝ ⎝ 3 ⎠ 3 ⎠⎦ ⎣ ⎡ I m sin(ωt ϕ ) ⎤ ⎢ ⎥ ⎢ I m sin ⎛ ωt − 2 π − ϕ ⎞ ⎥ ⎡ I m cos ϕ ⎤ ⎝ ⎠⎥ = ⎢ ×⎢ 3 s n ϕ ⎥⎦ ⎢ ⎥ ⎣ I m si ⎢ I sin ⎛ ωt + 2 π − ϕ ⎞ ⎥ ⎠ ⎥⎦ ⎢⎣ m ⎝ 3 (6) The reference signals (iiaref , ibref and icref ) of current tracking controlling can be obtained through the dq inverse transformation. The value of d in dq inverse transformation is the PI regulator output I′P, while the value of q is the reactive component (IQ) of load current. Both of them are as follows. sin ωt ⎡ ⎡iaref ⎤ ⎢ ⎛ ⎢ ⎥ ⎢sin ωt 2 π ⎞ ⎢ibref ⎥ = ⎢ ⎝ 3 ⎠ ⎢ ⎥ ⎢ ⎢⎣ icref ⎥⎦ ⎢sin ⎛ ωt + 2 π ⎞ ⎢⎣ ⎝ 3 ⎠

cos ωt ⎤ ⎥ 2 ⎞ ⎥ ⎡I ′ ⎤ ⎛ ωt − π p ⎝ 3 ⎠⎥⎢ ⎥ ⎥ ⎢⎣ IQ ⎥⎦ 2 ⎞ ⎛ cos ωt + π ⎥ ⎝ 3 ⎠ ⎥⎦

THE SIMULATION

(7)

The output current can be controlled in inverter circuit with PWM Tracking Control Technology. Its working principle is shown in Figure  4, where the reference signal i ref e (iiaref , ibref and icref ) can make the output compensation current of inverter circuit have the same value, phase and frequency with the load fundamental reactive current, and there is no fundamental reactive current in power current.

Figure 5.

29

The power supply system with STATCOM.

the three-phase load current (ial, ibl, icl) and its wavelet filtered current (iaf, ibf, icf). These two figures indicate that the filter works well. Figure 8 shows the threephase command current of STATCOM (iaref, ibref, icref) and its actual output three-phase current (iac, ibc, icc). Figure  9  shows the waveforms of three-phase supply voltage (uas, ubs, ucs) and current (ias, ibs, ics). Before 0.4 s, the fundamental reactive current is very small, so is the current generated by STATCOM. The supply current almost has no change compared with the load current. After 0.4 s, both the fundamental reactive current and the current generated by STATCOM are big, and the supply current has great changes compared with the load current. 6

Figure 8. The command current and its actual output current.

CONCLUSIONS

With the development of microelectronics technology, The DFACTS device, based on full-controlled devices and inverter technology, has been continuously improving its reliability and reducing its ongoing cost. So devices, such as STATCOM, APF, have broad application prospects. The wavelet filtering was applied in STATCOM while its

Figure 9.

The voltage and current of the source.

simulation was done. The simulation results show that the application works well. ACKNOWLEDGEMENT This work is supported by the training project of young key teachers of Anhui University (02303301). REFERENCES

Figure 6. The voltage of the public power supply point (PCC) before and after the filtering.

[1] Wang Zhao-an, Yang Jun, etc. Harmonic suppression and Reactive power compensation [M]. Beijing: China Machine PRESS, 1985. [2] Liu Wen-hua, Song Qiang, Zhang Dong-jiang, etc. Equivalent Tests of Links of 50 MVA STATCOM [J]. Proceedings of the CSEE, 2006, 26(12): 73–77. [3] Zheng Dong-run, Qiao Wei-dong, Liu W en-hua, etc. Field tests of 500 Mvar STATCOM [J]. East ChinaElectric Power, 2007, 35(1): 47–50. [4] Weon-Ki Yoon, etc. Reactive Power Measurement Using the Wavelet Transform [J]. IEEE Transactions on Instrumentation and Measurement (S0018–9456), 2000, 49(2): 246–252. [5] Zheng Chang_bao, etc. Reactive Power Measurement by Wavelet Transform and Hilbert Transform [J]., Journal of System Simulation 2005, 17(4): 822–824. [6] Jiang Qi-rong, Xie Xiao-rong, Chen Jiang-ye, etc. Power system parallel compensation [M]. Beijing: China Machine Press, 2004.

Figure 7. The current of the load before and after the filtering.

30

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Fractional-order differential application and research in pavement cracks image enhancement W. Jiang & Y.W. Liu School of Science, Chongqing Jiao Tong University, Chongqing, China

H. Zhang Institute of Civil Engineering and Construction, Chongqing Jiao Tong University, Chongqing, China

ABSTRACT: In order to sharpen the image edge features while enhancing texture detail, a new image enhancement method of pavement cracks is proposed by combining fractional-order differential theory with Prewitt operator effectively. The new method, compared with the existing image enhancement methods, retains the advantages of fractional-order differential effectively, which improve signal of high frequency components, and enhance the signal of the intermediate frequency components, and nonlinear keep signal very low frequency characteristics. Simulation results demonstrate that the new method can be obtained the effect of continuous variation, which not only enhanced image texture well, but also improved the edge enhancement effectively. Objectively, using image edge evaluation parameters of texture features such as information entropy and average gradient for quantitative analysis and experimental verification shows that the new method achieves the purpose of image enhancement. Keywords: 1

image enhancement; fractional-order partial differential; Prewitt operator; pavement cracks a hot topic, and some research has shown the fractional-order partial differential’s advantages in image enhancement. Pu Y.F.[5] introduced fractional-order differential into digital image processing and brought forward fractional-order differential masking and its algorithm. Simulation results showed that for image signal with rich texture information, fractional-order differential is superior to integer-order differential in enhancing texture details information. By analyzing the signal amplitude and frequency characteristics, Yang Z.Z.[6] found the fractional-order differential operation can improve image edge and texture details, and extract the edge information can avoid to produce larger noise at the same time, so as to improve the SNR of image processing. Fractional-order variation image denoising model is put forward[7]. The experimental results show that the fractional-order variation models are very effective in improving the peak signal-to-noise ratio and maintaining image texture details. However, to the best of our knowledge, there are few pavement crack image enhancement researches based on fractional-order differential theory, and almost fewer the study of pavement crack image enhancement, which combines the fractional-order differential with the Prewitt operator. Motivated by the above discussions, we

INTRODUCTION

The digital image processing technology, in the application of pavement crack detection, improves the automatization level of pavement crack recognition system based on the camera. In order to analyze the details of pavement crack effectively, there need to identify the enhanced image preprocessing. The existing crack detection algorithm is mainly based on three-dimensional terrain model of pavement crack automatic detection algorithm[1], artificial population algorithm[2], neural network algorithm[3], etc. All above mentioned are improved in a certain extent of regular crack detection algorithm, but for some slight and reticular cracks detection, these methods still cannot achieve the desired effect, and the computation is too large[4]. In digital image processing technology, it is not ideal when dealing with the edge texture details by using one-dimensional wavelet transform tensor product extension of two dimensional. Contourlet transform is due to the presence of the sampling process, which leads to the lack of translation invariance, the result of pavement image enhancement generates pseudo Gibbs distortion and the blurred crack edge. In recent years, the theory of fractional-order partial differential image processing has become

31

have combined fractional-order differential theory with Prewitt operator and applied them to pavement crack image enhancement; a new method of pavement crack image enhancement based on fractional-order partial differential is proposed. On the one hand, the fractional-order differential is good at retaining image texture details. Prewitt operator, on the other hand, is better to keep the edge character. The method inherits the advantages of both fractional-order differential and Prewitt operator, and it can enhance image details at the same time, also has a certain resistance to noise. Simulation results show that the new method can make obvious edge features, and smooth area information to retention. It can retain more texture details, thus improve the image visual clarity. So, it is more effective and practical on pavement crack image enhancement.

2

3

Assuming the gray function of image is F ( x, y ), the 3 3 pixel-domain is extracted, as follows. ( x , y + 1) ⎤ ⎡ F ( x , y 1) F ( x 1, y ) F (x ⎢ F ( x, y 1) F ( x , y) y ) F ( x, y + 1) ⎥⎥ ⎢ ⎢⎣ F ( x , y − 1) ( x + 11,, y ) F ( x 1, y 1) ⎥⎦ In order to simplify calculation, the differential is commonly used, instead of difference approximation, in the image processing. The operator template of horizontal gradient and vertical gradient of the Prewitt operator, as for the 3 × 3 area of discrete gray functions.

LH

DIFFERENTIAL EXPRESSION OF THE FRACTIONAL-ORDER PARTIAL DIFFERENTIAL

d f (t ) ( v )( v ≈ f tt)) + ( v f (t (t ) + 2 dtv Γ( v ) + + f t − n) n ! ( −v + n + 1)

)

⎡1 1 1⎤ ⎡1 0 −1⎤ ⎢ 0 0 0 ⎥ , L = ⎢1 0 −1⎥ ⎢ ⎥ V ⎢ ⎥ ⎢⎣ −1 −1 −1⎥⎦ ⎢⎣1 0 −1⎥⎦

∇fx = f ( x − , y − 1) + f x − 1,, y + f ( x − , y + 1) − f x + ,y −

The difference approximate expression of f tt) is exported based on the definition of fractionalorder differential[8]: v

NEW IMAGE ENHANCEMENT MODEL OF PAVEMENT CRACKS

f t

− f ( x + 1, y ) − f x + 1, y + 1)

(4)

∇f y = f ( x − , y − 1) + f x, y − + f ( x − , y − 1) − f x− , y +

)

− f ( x, y + ) − f x + 1, y + 1),

(5)

where ∇fx and ∇fy are horizontal gradient and vertical gradient, respectively. The step is Δx = 1, as the managed picture is digital image and the grayscale change of image is also limited, and the shortest distance of change is between two adjacent pixels.

(1)

The duration of the two-dimensional digital image on the axis is measured by pixel. In view of the digital image, the difference expression of the fractional partial differentials’ definition of the two-dimensional digital image f x, y ) is as follows:

∂ff x, y ) ⎧ ⎪⎪ f x, y f ( x 1, y ) = ∂x ⎨ ⎪ f x , y f ( x , y ) = ∂f x + , y ) ⎪⎩ ∂x

(6)

So the differentials form of ∇fx is as follows: ∂v f x , y ) ≈ f x, y ) + ( −v ) f x − 1, y ) ∂x v ( −v )( −v + ) + f x − , y) 2 Γ ( −v + 1) + + f ( x − n, y ) n!! ( −v + n + 1)

+ , y − 1) ∂f ∂f , y − ) ∂f + , y ) − − ∂x ∂x ∂x ∂f x, y ) ∂f ∂f + 1, y 1) ∂f ( , y + 1) − − − ∂x ∂x ∂x (7)

∇f x = −

(2)

In the same way the differential form of ∇fy is as follows:

∂v f x , y ) ≈ f x, y ) + ( −v ) f x, y − ) ∂yv ( −v )( −v + ) + f x, y − 2 ) 2 Γ( −v + 1) + + f ( x, y − n ) n!! ( −v + n + 1)

∂f

− , y + 1) ∂f ∂f − 1,, yy)) ∂f + , y + 1) − − ∂y ∂y ∂y ∂f + , y ) ∂f x, y + 1) ∂ff x, y ) − − − ∂y ∂y ∂y (8)

∇f y = −

(3)

32

∂f

The new model is as follows, though replacing the integer partial differential of the equations (7) and (8) with fractional partial differential.

( + ) ( − )(− )( + ) ( − ))(− ( + )⎤ ⎡ ( − ))(− − − ⎢− ⎥ 2 2 2 ⎢ ⎥ )( + ) ( − ))(( −v + 1) ( −v)( )( −v + 1) ⎥ ⎢ − ( − )(− +v − +v − +v ⎢ ⎥ 2 2 2 ⎢ ⎥ v −11 v −1 v −1 ⎢ ⎥ ⎢⎣ ⎥⎦ −1 −1 −1

+ , y − 1) ∂f ∂f v , y − ) ∂f v + , y ) − − v ∂x ∂x v ∂x v ∂f v x, y ) ∂f v + , y + 1) ∂f ∂f v , y + ) − − − v v ∂x ∂x ∂x v

∇fxv = −

∂f v

⎡ ( − )(− )( + ) ⎢− 2 ⎢ )( + ) ⎢ ( − )(− ⎢− 2 ⎢ )( + 1) ⎢ − ( − )(− ⎢ 2 ⎣

(9) ∂f

∇fyv =− −

v

∂f v

v

− , y + 1) ∂f ∂f − 1,, yy)) ∂f − − v ∂y ∂yv

v

+ , y + 1) ∂yv

+ 1, y ) ∂f v ( , y + 1) ∂f v ( x, y ) − − ∂yv ∂yv ∂yv

where ∇f ∇fxv and ∇f ∇f yv are the improved horizontal gradient and vertical gradient by fractional-order differential. To achieve a better image enhancement processing, choosing the first three of fractional partial differential difference expression (2) and (3) to form a 3 × 3 differential mask. In addition, the model of fractional-order horizontal gradient and vertical gradient the of Prewitt operator, which combining type (9) and (10), is as follows:

⎡ ⎢ 0 ⎢ ⎢v v 2 ⎢ ⎢ 2 ⎢v v 2 ⎢ ⎢ 2 ⎢v − v 2 ⎢ ⎢ 2 ⎢ ⎢ 0 ⎢⎣

∇fxv = v − )[ )[ f x, y + f ((x x, y − 1) + f x, y + )] − f x + , y − − f ( x + 1, y ) − f ( x , y + 1) )

4

∇f yv = v − )[ )[ f x, y + f (x− ( x , y ) + f x + 1, y )] ⎡ ( )( + ⎢− 2 ⎣ + f ( x, y )

− f ( x − 1, y + ) − f ( x )

⎤ + v ⎥ [ f ( x − 1,, y ⎦ f (x ( x + 1, y − 1)]

v −1 v −1

v v

2

v v

2 2

2

v v

2 7

2

0 ⎥

2 6

2

⎤

2

⎥

2 v v ⎥

⎥ ⎥ 2 2 2 7v v − 6 7 6 v −v ⎥ 4v 4 ⎥ 2 2 2 ⎥ 2 2 7 6 v −v ⎥ 2 2 ⎥ 3 2 3 2 2 2 ⎥ 2 2 2 ⎥ v −v v v v v 0 ⎥ 2 2 2 ⎥⎦ 3

2

2

3

2

2

EXPERIMENT AND RESULT ANALYSIS

This new model uses the Matlab7.1 for computer simulation experiment, and comparing with the existing part of the image enhancement method. The first set of experiments is to enhance image crack image 1, the experiment result is shown in Figure  1, where, (a) is the original crack image, (b)–(f) are enhance image results, which the new model take difference value of v . By the experiment result, if 0 0.7 , the experimental effect gets better with the increase of v ’s order gradually; if v > 0.7 , the experimental effect get worse with the increase of v ’s order gradually. The whole experiment gray became smaller with the increase of v ’s value; the enhanced image grey value is the most closely to the original image, so the effect is relatively good. when v = 0.65 . To illustrate, the new model of this paper has the advantages of the method of image enhancement. The second set of experiment uses an image crack

(11)

− f x, y +

⎤ − 1⎥ ⎥ ⎥ − 1⎥ ⎥ − 1⎥⎥ ⎦

v −1

Take example by the literature[5], rotating the mask on the up and down or so four directions centered on the point (x,y), then overlay, getting the new model gradient mask is as follows:

(10)

) ⎤ ⎡ ( )( + ⎢− + v⎥ [ f ( x − 1,, y 2 ⎣ ⎦ + f ( x , y) y ) f (x ( x − 1, y + 1)] ( −vv )( v + 1) − [ f x − 2, y − 1) 2 + f x − 2, y f ( x − 2, y) + 1]

( − )(− )( + ) +v 2 ( − )(− )( + ) − +v 2 ( − )(− )( + 1) − +v 2 −

, y + 1) )

( −vv )( v + 1) [ f x − 1, y − 2 ) 2 + f x, y f ( x + 1, y − 2) ] −

(12) The horizontal direction and vertical direction of gradient backward difference mask of the new improved model in is as follows, respectively:

33

Table 1.

Performance analysis.

Relative parameters

Information entropy

Average gradient

Original image Prewitt operator Fractional differential New method

7.0096 5.6614 7.8094 8.0853

7.1606 8.4880 12.3046 16.4117

image by the method in this paper achieved better effect in noise resistance and resolution than other methods. Thus, relative to the other methods, the new method in this article is an effective method of image enhancement.

Figure 1. (a) Crack image 1, (b) ν = 0.6, (c) ν = 0.65, (d) ν = 0.7, (e) ν = 0.75, (f) ν = 0.9.

5

CONCLUSIONS

Currently, seldom people combine fractionalorder differential theory with Prewitt operator to do image enhancement. A new pavement cracks image enhancement method based on fractional-order partial differential is proposed. The new method effectively inherits advantages, which keeps the characteristics of image edge and retains texture details of smoothing region. Simulation results show that the new method, compared with current image denoising methods, can not only suppress noise better, but also keep the characteristics of image edge better. Especially, it is better than current integer-order partial differential methods. It is an effective image denoising method. The result is approximatively expressed by fractional-order differential, so the future research is to determine a better differential order to obtain better results.

Figure 2. (a) Crack image 2, (b) new method, (c) fractional-order differential, (d) Laplacian operator, (e) Sobel operator, (f) Prewitt operator.

image 2 with rich texture details, the result is shown in Figure 2. It is not hard to see, the fractional-order differential treatment on texture details enhancement has certain effect. The exposure of image is too high by enhancement of the fractional-order differential and the visual effect is not very ideal. The visual effect of the enhanced image by Laplacian operator is not ideal as the measure in this article. The visual effect of the enhanced image by the Sobel operator and Prewitt operator appeared the phenomenon of distortion. After processing by the new model in this paper, the image not only has been enhanced in texture details, and the effect of the edge area enhancement is better compared to fractional differential. The enhanced image is improved in terms of signal-to-noise ratio and the average gradient compared with other strengthening methods through objective evaluation. Obviously, the enhanced

ACKNOWLEDGEMENTS The work is supported by the Research Programs for National Natural Science Foundation of China (11071266), the Natural Science Foundation of Chongqing (CSTC, 2012JJA1164), and the Technology Project of Chongqing Municipal Education Commission (KJ120401). REFERENCES [1] Tang, L. et  al. 2008. Automated Pavement Crack Detection Based On Image 3D Terrain Model. Computer Engineering 34(5): 21–38. [2] Zhang, H.G. et al. 2005. Pavement Distress Detection Based on Artificial Population. Journal of Nanjing University of Science and Technology 29(4): 389–393.

34

[3] Bray, J. et al. 2006. A neural network based technique for automatic classification of road cracks. Proceedings of International Joint Conference on Neural Networks 907–912. [4] Ma, C.X. et  al. 2009. Pavement Cracks Detection Based on NSCT and Morphology. Journal of Computer-Aided Design and Computer Graphics 21(12): 1761–1767. [5] Pu, Y.F. 2007. Application of fractional differential approach to digital image processing. Journal of Sichuan University (Engineering Science Edition) 39(3): 124–132. [6] Yang, Z.Z. 2006. Fractional order differential in the study of the application of the modern signal analysis and processing. Chengdou: Sichuan university. [7] Thangavel, K. & Karnan1, M. 2007. Automatic detection of asymmetries in using genetic algorithm. International Journal on Computer Methods and Programs in Biomedicine 87: 12–20.

[8] Pincherle, I. 1990. Fractional differential equations. 20–23. SanDiego: Academic Press. [9] Jiang, W. 2011. New image denoising model based on fractional-order partial differential equation. Journal of Computer Applications 31(3): 753–756. [10] Guo, L.Z. & Zhao, J.H. 2007. Edge detection based on wavelet transforms. Journal of Qingdao Technological University 28(2): 78–80. [11] Zhang, J. et  al. 2009. Automatic identification of pavement crack image enhancement technique. Journal of China and Foreign Highway 29(4): 301–305. [12] Yang, Z.Z. et  al. 2008. Edge detection based on fractional differential. Journal of Sichuan University (Engineering Science Edition) 40(1): 152–157.

35

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Molecular algorithm in solving the shortest path problem of the application and research L. Zhang College of Sciences, Chongqing Jiaotong University, Chongqing, China

ABSTRACT: Interest in DNA computing has increased overwhelmingly since Adleman successfully demonstrated its capability to solve Hamiltonian Path Problem (HPP). One line of DNA computing research focuses on parallel search algorithms, which can be used to solve many optimization problems. In this article, we have implemented a new DNA computation-based optimal path planning algorithm on 3D space. The molecular programming algorithm imitates the biological evolution mechanism through artificial programming to enhance the opportunities for searching the shortest moving path while avoiding obstacles. Simulation results are presented to show the effectiveness and applicability of the proposed approach. Keywords: 1

DNA computation; optimal path planning; crossover; mutation

INTRODUCTION

evolutionary process. This work presents clear evidence of the ability of molecular computing to solve the NP-complete problem with mathematical operations. The article is organized as follows: Section  2 introduces the DNA computing in detail. Section 3 introduces the DNA program to solve the optimal path planning problem in three-dimensional (3D) space. In Section 4, experimental results by simulated DNA computing are given. Conclusions and discussion are presented in Section 5.

The shortest path between two given points over a curved surface is a difficult problem in theoretical and practical field. It is widely used in the mapping, the span of electric power and LOC, the layout of the railway, the road, the path of robot etc. It will take huge economical benefit if it has been resolved well. Up to now, genetic algorithm and evolutionary algorithm have been widely applied for system optimization and industrial applications. Because traditional GA or EA can neither represent the diverse genetic information nor can better imitate the regulation of genes to the genetic processes, some biological operations at the gene level cannot be effectively presented in the existing GA or EA. Especially on three-dimensional (3D) space it is a thorny issue. In addition to GA and EA, DNA computing [1] has recently captured more and more attention [2] since DNA sequences encode plentiful genetic information. This article presents an optimal path planning method for the moving object on threedimensional (3D) space using DNA molecular structures and molecular programming. The algorithm is designed to analogously accomplish the evolutionary process. In the proposed method, the 3D workspace is inserted into several points, the DNA sequences are used to represent the points, connecting all optimal points while avoiding the obstacle areas builds the optimal moving path, and the shortest path for obstacle avoidance in the workspace is finely characterized with DNA

2

DNA COMPUTING

In the late 1950s, Feynman first introduced the idea of computation at a molecular level, but his idea was not implemented by experiment for a few decades. In 1994, Adleman wrote the first article that demonstrated that DNA strands could be applied for dealing with solutions of the NP-complete Hamiltonian Path Problem (HPP). Since then, the possibility of DNA computation has attracted many researchers’ attention. In 1995, Lipton [3] wrote the second article that showed that the new techniques could also be used to solve the NPcomplete satisfiability (SAT) problem. Adleman and his co-authors [4] proposed sticker for enhancing the Adleman–Lipton model in 1999. DNA computation uses chromosomes of DNA to describe a specific problem and to manipulate the chromosomes using techniques commonly available in molecular bank for simulating

37

the goal along a shortest path while avoiding static obstacles on the workspace. In practical projects, this is very common. In this article, the algorithms will bypass the designated area, and will access a feasible and shortest path successfully. 3.1

Figure 1. The flow of the optimal algorithm.

operations, which can be used to seek for the solution. The central idea of DNA computation is the Watson–Crick model of DNA structure, which specifies complementary binding properties of DNA molecules. The basic elements of biological DNA are nucleotides. Due to their different chemical structure, nucleotides can be classified as four bases: A (Adenine), G (Guanine), C (Cytosine) and T (Thymine). A single strand of DNA can be likened to a string consisting of a combination of the above-mentioned four different symbols; this means that we have an alphabet ∑ {A, G, C, T} to encode information. Any single-stranded DNA will adhere tightly to its complementary strand, in which G always pairs with C and A always pairs with T, and vice versa. The process of DNA computation can be refined into two steps. The first step is to generate all possible solutions to the problem by mixing DNA solutions. DNA complementary binding reactions occur in parallel and extremely fast upon mixing. The second step is to isolate correct solutions through repeated separations of the DNA strands from incorrect solutions and potentially good solutions. A schematic representation of DNA computation is given in Figure 1. DNA computation is attractive mainly for three reasons. First, the computation realizes fast parallel information processing. Second, the process is remarkably energy efficient. Finally, DNA molecules have very high storing capacity. DNA computers have been shown to be at least equivalent to a classical Turing machine [5]. 3

The initial path and coding schemes

The initial DNA sequences are generated randomly, suppose the workspace is designated. The origin is A( x0 , y0 ), the goal is B ( xn , yn ), wij is the edge weight between two random nodes i and j which have a connection with each other. Divide the area of [ 0 , xn ] into n slices, meanwhile the partition of [ y0 , yn ] is ascertained. Each point we use the DNA strands to encode, the length is 20, and mark it as Oi (i , 2, , n ) . Edges that connect nodes encoded and formed by the following sequence: the first part is a 10 codes comes from the 3′ of the Oi; the middle part is the codes of the edge weight; and the last 10 codes appears at the 5′ of the Oj. In Figure 2, let O2 O3 → O4 as an example to explain the coding of the paths. 3.2

Choose the fitness function

When listing facts use either the style tag List signs or the style tag List numbers. 3.2.1 The design of avoiding obstacles The slope and length are two important factors in path planning. In the numerical simulation example, the value of z on the roadblock will be added a large constants cˆ ( ˆ max( ij )) , zij is the elevation value of the corresponding point which is decided by the landform. Because the fitness of the slope and length are all poor, the individual which gets across the obstacle areas will be eliminated during the operation. 3.2.2 The calculation of fitness value After times of genetic operation, the new path set is appearing, we define it for Γ k , and suppose rki is the node sequence of the ith path in Γ k .

PATH PLANNING AND OBSTACLES AVOIDANCE

Our goal is to characterize a feasible moving path which is required to start from the origin and reach

Figure 2.

38

The coding of paths.

rki

In the path searching, in order to obtain some better individual, part of the initial DNA sequences should be interchanged, the via-points in an adjacent area can be displaced by each other through the crossover process to change their forward directions and have more chances to avoid the obstacles. Crossover operation can avoid the algorithm getting into the local optimal solutions, the design of the operator is especially important. The descendant must keep the excellent ingredient of their fathers, and must have their validity.

i i i {( xki1, yki 1 zki1 ),(x ) (xki 2 yki 2 , zki 2 ), ,( xkn , ykn , zkn )}

(1) In which, i 1, 2, …, m , m is the quantity of the g, and g is the evolutional genroutes; k erations. Suppose Lik is the length of the ith path. Lik

rki

n−1

= ∑ ( xki

j+ j )

i 2 xkj ) + ( yki

i 2 j ) − ykj ) j+

+ ( zki

i 2 jj+ ) − zkj )

j =1

(2)

3.3.2 Mutation Any change in a DNA sequence is called a mutation, it includes point mutations and structure mutations. In mutation operation, the randomly selected bit in the sequence is replaced by the rule that A changes to T and vice versa. To avoid more duplicate points generated by crossover, the mutation has function to enhance the chances of exchange between the points in a small scope. For the optimal path searching, let the vij and v( i )()( j ) replace the vi( i ) and v j ( j ) when wij + w(ii j + ) < wi ( i + ) + w j(j ( j + ) . This algorithm can eliminate overlapping phenomenon of the sides. When we improve the route by mutation, to every point, their correlative sides are all taking place around it. So we only have to carry through the optimized computation between the point and the points of their adjacent array to improve the loop. This method can enhance the efficiency, and decrease the searching space. Before the mutation:

The Dki

is the sum of variance, the distant height of two connecting nodes will be measured by it. E = (z j

n−1

+ z j + z jj++1 ) 3

(

Dki = ∑ ( zki j =2

jj+ )

i E )2 + ((zzkj

E )2 + ( zki

j− )

)

E )2 3 (3)

So the fitness function is Fitnessk (i ) = 1000 − 0 * Lik

− 0.. * Dki

i 1, 2, …, m; k 3.3

g

(4)

The manipulation of DNA computing

After the initial DNA sequences are generated, we should reproduce the route by the adaptive value which was determined preliminary by the length of the path. The purpose of reproduction is to make the individual which has bigger adaptive value has more chance to breed the descendant. This article adopts the ways of fitness rule and distillate conservation to reproduce.

⋯GACT ATATCGCGGGTTCAACGTGC A GACG ⋯

After the mutation: ⋯GACT ATATCGCGGGCAGTTCGTGC A TAGA ⋯

3.3.1 Crossover Crossover is a process of exchanging genetic information between two DNA sequences. For simplicity, the one-point crossover is adopted here where the crossover point is determined uniformly at random. However, a multiple point crossover is also permissible. After crossover, the codes placed in front of the crossover point are kept invariant, while the subsequent DNA codes are interchanged. Before the crossover:

3.3.3 The result After the crossover and mutation, there will be appearing a series of new generation paths, if the fitness is not satisfactory, we should repeat the above operations, all of the operations will make the individual’s fitness value and the average value becomes higher and higher. Until the best individual’s fitness is achieving a restrict value or the individual’s and the average fitness couldn’t change anymore, then the iterative process can be converged to the optimal solution, and the algorithm finished. The final process is to separate the DNA strands. Gel electrophoresis is a technique for separating DNA strands according to its length through a gel in an electrical field based on the fact that DNA is negatively charged [6]. As the separation process

⋯ GACT A ATATCGCGGG TATCGCGGG TTCAACGTGC GACG ⋯ ⋯ ATAG CAGCTCATCG C CAGTTGACAT TCTG G⋯

After the crossover: ⋯ GACT A ATATCGCGGG TATCGCGGG CAGTTGACAT TCTG ⋯ ⋯ ATAG CAGCTCATCG C TTCAACGTGC GACG G⋯

39

Table 1.

Figure 3. Gel electrophoresis process.

continues the separation between the larger and smaller chains increases as depicted in Figure  3 [7]. At last, we should incise the joint between two nodes by cutting enzyme, a series of polynucleotide fragments are received. Then the path nodes are decided by contradistinguishing the initial coding schemes. 4

n

1

2

3

4

5

6

7

8

xc yc p q u v h

25 87 1.5 1.5 30 20 57

64 28 2 3.5 65 12 19

34 53 3 2 21 13 −32

77 92 2 2 6 25 38

94 55 3 3 11 14 24

65 24 1.5 1.5 13 9 −22

15 11 3 2 11 13 −14

93 14 3 3 13 11 −17

Table 2.

DEMONSTRATIONS AND VERIFICATIONS

This article cites the simulation landform of the literature [8] to search the square space on 100*100. we use the analytical function to simulate the original terrain, and the curved surface has been constructed by the following function:

Parameters of simulated altitude.

Coordination of result nodes.

n

1

2

3

4

5

x y z

5 92 29.2

15 63 2.2

25 54 −23.4

35 43 −14.3

45 19 8.4

n

6

7

8

9

10

x y z

55 12 −1.2

65 13 −3.6

75 10 −2.45

85 9 −12.5

95 9 −15.4

pi q ⎧ ⎛ ⎛ y yc ⎞ i ⎫⎪ ⎪ x xci ⎞ i ∑ hi exp ⎨⎪− ⎜⎝ u ⎟⎠ − ⎜⎝ v ⎟⎠ ⎬⎪ i i i =1 ⎩ ⎭ n

z ( x, y

(5) The data in Table  1 are the parameters of the above-mentioned function. In the simulation example, suppose the initial positions of the object and goal are specified, respectively, at (5,92) and (95,9). The area of x is established for [5,95] and y is been decided in [9, 92]. Then take 10 as the length of stride to divide the area of x into nine segments. We can obtain the value of z through the curved surface function and the value of the x and y, the raised square areas are the roadblock. In the example, choose eight points to divide the workspace into nine slices, every point denoted as the feasible via point, the optimal moving path is obtained by connecting the best via point which is determined by the applying the proposed computing algorithm with the fitness function. After specifying the roadblock, the DNA computing-based search algorithm is activated in the path planner to generate via points for the short and safety path to the goal. The scale of the initial population is 20, let 0.1 as the reproductive probability, 0.7 is the crossover probability, and the mutation probability is 0.05, the most iteration number is three hundreds, and cˆ = 500 . After the simulation example, the length of the best path is L = 190.87. But the

Figure 4.

The three-dimensional sketch drawing.

length when it is only been connected by beeline between two points is = 201.2 , the result is better than the method mentioned in [8], the optimization efficiency λ is:

λ=

S−L × 100% = 5.13% S

(6)

The best node sequences of the function are presented in Table 2. Simulation results confirm the feasibility of our proposed approach (see Figs. 4 and 5). It is obvious that with this method the path planning is reasonable, it bypasses the roadblock and some steep places well.

40

nature. Hence, the applicability of DNA computing could be extended into greater fields of other engineering-related problems. REFERENCES

Figure 5.

5

1. Adeleman L.M., 1994, Molecular computation of solutions to combinatorial problems. Science. (266), 1021–1024. 2. Garzon M.H. and Deaton R.J., 1999, Bimolecular computing and programming, IEEE Trans. On Evolutionary Computation, vol. 3, no. 3, 236–250. 3. Lipton R.J., 1995, DNA solution of hard computational problems. Science. (268), 542–545. 4. Roweis S., Winfree E., Burgoyne R., Adeleman L.M., et  al., 1999, A sticker based model for DNA computation. Second Annual Workshop on DNA Computing, Princeton University. American Mathematical Society, 1–29. 5. Maley C.C., 1998, DNA computation: theory, practice, and prospects. Evol. Comput, 6(3), 201–229. 6. Paun G., Rozenberg G., Salomaa A., 1998, DNA Computing: New Computing Paradigms. Lecture Notes in Computer Science, Springer-Verlag, Vol. 1644 106–118. 7. Amos M., 1997, DNA Computation. PhD Thesis, The University of Warwick, UK. 8. Huang Zhang-can, Chen Si-duo, Kang Li-shan. 2000, Solving the Shortest Path on Curved Surface Based on Simulated Annealing Algorithm. The transaction of Wuhan University, 6(46)3: 273–276.

The contour schematic drawing.

CONCLUSION

It is expected from experimental results that the shortest DNA sequence length will represent the required optimal path planning on 3D space. Based on the DNA computing algorithm, a novel process for charactering the optimal path while avoiding obstacles is developed. With the successful confirmation of the expected result, the applicability of DNA computing could be extended into many more complex problems of this type of

41

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

A method about video quality assessment in video phone service over 3G network C. Feng & L.F. Huang School of Information Science and Technology, Xiamen University, Xiamen, China

W.J. Xu Jimei University, Fujian, China

ABSTRACT: The method about video phone service and quality evaluation on TD-SCDMA has been studied in this article. A wireless module LC6311+ is used as the front-end video data transceiver equipment and the ITU-T J.247 PEVQ algorithm is used as the back-end video quality evaluation model. The LC6311+ wireless module is developed via AT commands over videophone CS64k link. After sending a prepared video compression stream over the CS64k link, the receiving end will get the video stream and preserve it. The original video sequences and the degraded video sequences can be obtained after decoding them. The degraded video sequences are imported to PEVQ model, comparing them with original video and analyzing them. The score of the video quality can be obtained. The results show that this method may be effective for an objective assessment of the video quality of the 3G network evaluation. It is useful for mobile network operators to optimize scalable video quality. Keywords: 1

in video quality assessment; TD-SCDMA; video phone; PEVQ; LC6311+

INTRODUCTION

the front-end for capturing the video data and the back-end for video quality assessment [2]. The LC6311+ wireless module is described in this paper. A standard H.263 video stream is sent by videophone services through the CS64k channel to the receiver. It is studied by the degraded video stream. After obtaining the decoded source video and the degraded video, the video quality evaluation score can be calculated from the PEVQ model. The main indicator of PEVQ is introduced in the final part [3].

In 3G or LTE mobile communication system, the most attractive service is about video transmission. But the key indicator that affects the user experience is the wireless video quality. The methods of currently wireless video quality assessment are divided into two basic categories: Subjective evaluation and objective evaluation. Many observers are required to rate the quality of the video in subjective evaluation. The disadvantage is time-consuming and expensive. So it is not widely used. It is widely used for objective evaluation of video quality because it has less cost, time-saving, objective, reliable and repeatable etc. The common algorithm about image quality assessment is the Peak Signal-To-Noise Ratio (PSNR), the edge of the Peak Signal-To-Noise Ratio (EPSNR) and structural similarity (SSIM) [1]. There is a number of recommendations for video quality assessment including ITU-T J.247 recommended evaluation model—Perceptual Evaluation of Video Quality (PEVQ). It has become one of the major video quality assessment methods. The PEVQ is the full reference video quality evaluation model. A complete reference source video is needed. For evaluating the video quality of videophone service, there are two main steps:

2

THE BASIC PRINCIPLE OF VIDEOPHONE SERVICE QUALITY EVALUATION

As the current TD-SCDMA test terminal does not provide relevant programming interface, the TD-SCDMA test terminal cannot be directly used to get the video data. The video sample data files saved on the computer’s hard disk are encoded in accordance with videophone 3G-324M protocol. Then the encoded video stream is transmitted over videophone link (CS64k) to simulate videophone data transmission process. The 3G-324M protocol is integrated on the LC6311+ wireless module. Meanwhile, the video quality assessment method

43

and memory system includes the core processor which is a dual-core processor: BlackFin DSP core, running physical layer protocol and AMR codec; ARM926 MCU core, running senior protocols, and communication with external interfaces. External interface provides external connectivity including power input, communication interface, control signals and debugging interface. The LC6311+ module uses the USB interface to communicate with the PC. It provides two Endpoint links to achieve CS64k signaling and data interaction which is operated by AT commands. Endpoint6 is used as the link1 which mainly makes video calls and hangs up the phone. Endpoint4 is used as link2 for the interaction of video data. All the USB endpoints are developed as virtual serial port. In a video call, the sending data rate is required to meet channel rate CS64k, i.e. 64k = 48k (for video data) +12.2 k (for voice and data)  + control information. Because the voice data are acquired with the LC6311+ module, the control information is also packaged in it. The terminal application need to control the transmission rate of the video data. The LC6311 + modules are operated via AT command set which mainly reference the 3GPP 27.007, 3GPP 27.005 and ITU-T V.25. The AT commands is generally used in connection with the communication between the terminal device and the PC [5]. The flowchart of operation with LC6311+ including the module initialization, startup, calling a video phone and shutdown is shown in Figure 2.

Figure  1. Block diagram of video quality assessment for video phone.

uses the ITU-T J.247 PEVQ (Perceptual Evaluation of Video Quality). The algorithm compares source video (SRC) and the video after transmission (PVS). Then the MOS score which is an objective assessment to simulate the user’s subjective evaluation can be calculated [4]. The basic principle of the design is shown in Figure 1. 3

THE VIDEO PHONE SERVICE SIMULATION WITH LC6311+

We use the LC6311+ TD-SCDMA wireless module to call a CS64k videophone service. The video data is transmitted through the channel so as to obtain the test video data. The terminal uses MTK Co.  Ltd ASIC chipset, including RF subsystem and analog baseband subsystem. The RF subsystem includes radio frequency transceivers, amplifiers and transceivers, such as switches and filters. The analog baseband subsystem integrated DAC, PMU. PMU (Power Management Unit) module contains a number of different voltage and current path in the power supply unit to meet the needs of the power supply. Digital baseband processor

Figure 2.

44

Call a video phone with LC6311+.

luminance part and the chrominance part of the edge images. E x [ i , j ,t ] =

Pedge,x [i, j ,t ] − Sedge,x [i, j ,t ] Sedge,x [i, j ,t ] ( x isY i oor Cb or Cr)

W −1 H −1

Figure 3.

The saved SRC and PVS.

Ex [t ] = 2

In Figure  2, the OpenComm() is used to open the serial port (USB Endpoint6) and configure it. Init_LC() function is used to initialize the module. After initialization the Start() function is used to power the module. VoiceDial() function is used to make voice calls and VideoPhone() function is used to make video calls; The Hand_up() function is used to hang up a voice call and Hand_up_VP() function is used to hang up the video phone; and Off() function is used to execute the shutdown command. When the wireless module is powered on, the PC executes module initialization, startup, make video calls and other AT command operation. The compressed bit stream which is sent by wireless module can then be decoded by the PC. When we get the transmitted video data, they are imported to the PEVQ module. After calculating, the video quality MOS score can be obtained. The sample frame of SRC and PVS is shown in Figure 3. 4

2

E x [ i , j , t ] w[ i , j ]

i =0 j =0 W −1 H −1

∑ ∑ w [i j ] i =0 j =0

( x isY or Cb or Cr) w[ i , j ]

⎛ i ⎞ ⎛ j⎞ sin π ⋅ sin π ⎝ H⎠ ⎝ W⎠

(2) (3)

The H and W represent the row and column. The i and j represent the coordinate of the pixel. 4.1.1 Luminance indicator When the x is Y, it means equations (1) and equation (2) compute the luminance indicator. The indicator not only indicates loss of sharpness. Also, the introduction of sharpness is registered as a distortion. This indicator is perceived as a loss of sharpness. Some source sequences have an overall higher edginess than others. The introduction of edginess in areas with a lot of edginess is less disturbing than the introduction of sharpness where little edginess is originally present. The luminance indicator is then calculated by averaging the frame wise edginess distortions over time

THE VIDEO QUALITY ASSESSMENT WITH PEVQ

In this article, PEVQ algorithm is used to evaluate the video quality of 3G videophone. PEVQ is developed by OPTICOM and recommended by the ITU-T J.247. The evaluation results have been recognized by many manufacturers. It gives different weighting coefficients for different indicators in order to approximate subjective evaluation value [6]. The PEVQ is a full reference model which requires the same resolution source video sequence (SRC) and test video sequence (PVS). The PEVQ uses several fitting parameters which are extracted from the time domain, space domain, luminance and chrominance domain of the SRC and PVS. After extracting the parameters, they are imported to the model. So the MOS score can be obtained. All the indicators are described as follows: 4.1

∑∑

(1)

Lum IIndicator =

1 N −1 ∑ EY [t ] N t =0

(4)

4.1.2 Chrominance indicator The chrominance indicator uses a similar approach as the luminance indicator. When the variable x in equation (1) and equation (2) represents the Cb or Cr. The Ecb [i, j ,t ], Ecr [i, j ,t ], Ecb [t ] and Ecr [t ] are calculated separately. As for the deviation signal, the maximum of the color saturation of the reference signal and the degraded signal is taken as above. The change in edginess of both color components is evaluated. The extraction methods of chromaticity component and luminance component are similar. Chrominance value contains components of Cb and Cr which are calculated separately as above. The chrominance indicator is then calculated by averaging the frame-wise edginess distortions over time:

Spatial variability indicators

The spatial variability indicators involve two indicators: the luminance indicator and the chrominance indicator. They are calculated based on the

Chrom Indicato I r=

45

1 N −1 ∑ (Ecb [t ] + Ecr [t ]) 2N t = 0

(5)

4.2

Table 1. Weight values and model parameter (QCIF).

Temporal variability indicators

The edginess indicator is a pure spatial indicator. However, the spatial content of a sequence is judged more critically in case of still images than for images with fast motion and rapid changes. To reflect this, we introduced two indicators: The Omitted Component Indicator (OCI) and the Introduced Component Indicator (ICI). The temporal variability of the processed video signal is also influenced by transmission errors and the presence or absence of frame repeats. As a result, the temporal variability is best measured on the luminance of the source sequence. d [i, j t ] = SY [i j t ] − SY [i, j t −

dO/I [t ] = 5

H −1W −1

W H

i =0 j =0

∑ ∑d

5

/ + [i,

1 N −1 ∑ d [t ] N t =0 O

]

1 N −1 2 ∑ dI [t ] N t =0

β[i]

0 1 2 3

LumIndicator ChromIndicator OCI ICI

6.184 2.501 −5699.585 −0.866

0.168 1.351 0.0036 0.311

−1.449 −17.748 8.729 −6.767

The VQA result of PEVQ method.

j ,t ]

Hall

Carphone Suzie

2.9136 2.9133

Akiyo

2.7821 2.7665

(6) 4.3

The aggregation of indicators

The perceived video quality is estimated by mapping the Indicators to a single number using a sigmoid approach. Let I[i] represent the indicators. Then the mapping function may be defined by a set of input scaling factors I[i], a set of scaling factors w [i ], α [i ] and β [i ]: 3

Score Offs O et + ∑ i =0

w[ i ] 1 + eα [ i ] I [i ] β [i ]

(10)

Mapping coefficients used for QCIF resolution. The offset is equal to −0.93 [6]. The PEVQ Score of five test video sequences: Foreman, Hall, Carphone, Suzie and Akiyo are given in Table 2. 5

CONCLUSION

With the rapid development of 3G video phones and other wireless video services, in order to enable users to obtain a better subjective experience, the importance of video quality assessment in the development of wireless video services, network optimization and other areas are also increasingly prominent. In this article, LC6311+ wireless module and FFmpeg video codec library use the LC6311 modules via serial programming completed encoding YUV material sent and program received design; realize the video of the original material and synchronization obtained through material wireless transmission after the; adoption PEVQ objective evaluation of the degradation of the video quality of the video. After the experiment, the method can be effectively applied to TD-SCDMA system CS64k videophone business objective evaluation of video quality, the video can optimize the quality of service operators, and network planning and optimization work to provide a convenient way.

(7)

(8)

The Introduced Component Indicator (ICI) is then calculated using an L2 Norm over the framewise introduced distortions over time ICI C =

α[i]

PEVQ 2.8881

The Omitted Component Indicator (OCI) is then calculated by averaging the frame-wise omitted distortions over time. OCI C =

w[i]

Foreman

Frame loss caused by wireless transmission not only causes big difference between the two continuous frames in SRC and PVS, but also significantly influences the subjective feeling ability of human eyes. The temporal variability indicators in PEVQ model is defined as equation (6), where d [i, j ,t ] is computed per pixel per frame, SY [i, j ,t ] and PY [i, j ,t ] are representing the luminance of pixel at the SRC and PVS of frame t. Therefore, this algorithm can only be used in full reference evaluation model of video quality. The HVS reacts differently if a new component is introduced to the signal then if it is removed from the signal. Therefore, two different indicators are evaluated. To measure the omitted component part, the negative part (d [i, j ,t ] ) and the positive part − (d [i, j ,t ]) of d [i, j ,t ] are separately calculated as + equation (7). 1

I[i]

Table 2.

]

− PY [i j t ] − PY [i, j t −

i

(9)

46

ACKNOWLEDGEMENTS

[2] Margaret H Pinson and Stephen Wolf, A New Standardized Method for Objectively. IEEE Transactions on Broadcasting, 2004, 50(3):312–322. [3] Chao Feng, Image quality assessment-oriented frame capture of video phone over 3G system; Anti-counterfeiting, Security and Identification 2010, Page(s): 359–362. [4] Xin-Bo Gao, Quality Assessment Methods for Visual Information, Xi’an, XiDian University Publishing house, 2010. [5] Leadcore Technology Co. Ltd. The Video Phone Development Manual_1.10 based on LC6311+ [DB]. [6] ITU-T J.247 Objective perceptual multimedia video quality measurement in the presence of a full reference[S], 2008.

The work presented in this paper was partially supported by 2011  National Natural Science Foundation of China (Grant number 61172097), 2014  National Natural Science Foundation of China (Grant number 61371081), 2012  Natural Science Foundation of Fujian (Grant number 2012  J01424) and Foundation by Comba Co., Ltd. REFERENCES [1] Kjell Brunnstrom. David Hands. VQEG Validation and ITU Standardization of Objective Perceptual Video Quality Metrices[J]. IEEE Signal Processing Magazine [96] MAY 2009.

47

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Analysis on the degree of the international fragment of China’s manufacturing industry based on processing trade Y.H. Yang School of Economics and Management, Yunnan Normal University, Kunming, China

ABSTRACT: This article constructs measuring indexes of the international fragmentation production of manufacturing industry based on the data of processing trade. Using China’s manufacturing trade data encoded into 4-digit HS code, it measures the international fragmentation production of China’s 26  manufacturing industries. The result shows that there are industrial differences in the participation of international fragmentation of China’s manufacturing production, that is, the degree of international fragmentation production of some industries is high, while others’ is low. Some industries participate in the international fragment of production mainly through the processing trade exports. While others participate in the international fragment of production mainly through the processing trade imports. Other participations of imports and exports are more prominent. Keywords: international fragmentation of production; processing trade; China’s manufacturing industry 1

INTRODUCTION

extent of our participation in the international fragmentation of production has been increasing. Manufactured goods are the subject of China’s participation in the international fragmentation of production, which is an important form of China’s participation in international economic activity. It is a valuable research to accurately measure the degree of international fragmentation of Chinese manufacturing industry and to comprehensively analyze China’s manufacturing industries’ participation in the international fragmentation of production. Input–output method has been commonly used in the literature to measure international degree of fragmentation in the production of our manufacturing industry (Zhao lei and Yang Yonghua, 2011). Due to the unavailability of the key variable—the intermediate input amount of data—the intermediate goods imports were assumed, that is, the proportion of industry’s total import of intermediate products is equal to the import of k industries. There are some deviations from the intermediate input the actual situation of various industries. Therefore, the input–output method to measure the true degree of international fragmentation of China’s manufacturing production has its limitations. Therefore, this article uses the data of processing trade which is relatively easily available to measure China’s manufacturing industries’ international fragmentation of production.

Since the reform and opening up, China’s export expensed rapidly. The rank of the total amount of China’s export rose from the 28th in 1980 to the 2nd in 2007. From 2009 to 2011, China remains the world’s largest exporter. China’s exports are mainly manufactured goods-oriented. In recent years, the share of manufactured goods in total exports has remained at 95%, and exports of manufactured goods grew faster than the growth rate of exports of all goods. The basic phenomena of current products is that different processes, sectors, and components in the production process scattered to different countries and regions, thus the same final product was manufactured by different countries or regions’ participations in the international fragmentation of production. Two or more countries or regions completed a final production through the Foreign Direct Investment (FDI) or outsourcing activities to produce or assemble parts. A distinguishing feature is that the intermediate input circulated in nations, resulting in a large number of intermediate goods trade. Processing trade is one of the basic forms of the international fragmentation of production, as well as China’s main way to participate in the international fragmentation of production (Lufeng, 2004). China’s processing trade export share of total export increased from 5% in 1981 to 56% in 1996, then at around 55% by 2009. Obviously, the

49

2

MEASUREMENTS OF INTERNATIONAL FRAGMENTATION IN PRODUCTION BASED ON PROCESSING TRADE DATA

in 1997 and in 2002, other manufacturing industry in 1997, other industry in 2002, arts and crafts industry and the processing of waste recycling in 2007. We categorized some industries and unified their names, such as classifying agro-food processing and food manufacturing as food processing and manufacturing, classifying wood processing and wood, bamboo, cane, palm, grass products industry and furniture industry in the year of 2002 and 2007 as wood processing and furniture manufacturing, classifying sawn timber processing and furniture, wood and man-made board manufacturing and grass bamboo rattan palm products manufacturing in 1997 as wood processing and furniture manufacturing, so as to get 26 manufacturing industries. For the convenience of following analysis, we encode the 26 manufacturing industries (see the note of Table 1).

Processing trade is to use foreign parts, components and other resources and to process, manufacture, assemble and produce in domestic country, which is then sold in foreign trade. Processing trade is one of the basic forms of the international fragment of production, as well as China’s main way to participate in the international fragment of production. Measurements of international fragmentation in production based on processing trade data are mainly used in many literatures (Yeats, 1998; Görg, 2000; Baldone et al., 2001, 2007; Helg and Tajoli, 2005). Referenced Balassa index of revealed comparative advantage, and Baldone et  al. (2007), we can construct a metric index of China’s manufacturing industries international fragment of production as following: RFP PijT = RFP PijM =

PT T jt / T jt PT Tt / Tt

, RFP PijX =

PTM M jt / TM M jt PTM Mt / TMt

PTX X jt / TX X jt PTX X t / TX t

Table 1. Tendency of international fragment of China’s manufacturing production based on the measurement of processing trade export from 2001 to 2008.

,

Industry year C1 Average

(1)

Average

3

C3

C4

C5

C6

C7

0.55 0.14 0.05 0.48 0.66 0.75 0.67

Industry year C8

Here, RFP PijT , RFP PijX and FP PijM , respectively, represent the indicators of international fragment of manufacturing production measured from the perspective of trade import and export volume, export and import; i represents trading nation, and j represents some industry. PT T jt and PT Tt , respectively, represent j industry’s import and export volume of processing trade, and Chinese manufacturing industry’s import and export volume of processing trade in the t year. T jt and Tt , respectively, represent j industry’s import and export volume, and Chinese manufacturing industry’s import and export volume in the t year. PTM M jt , PTX X jt , PTM Mt and PTX X t , respectively, represent j industry’s import and export volume of processing trade, and Chinese manufacturing industry’s import and export volume of processing trade in the t year. MT T jt , XT T jt , MT Tt and XT Tt , respectively, represent j industry’s import and export volume, and Chinese manufacturing industry’s import and export volume in the t year.

C2

C9

C10 C11 C12 C13 C14

1.02 1.35 1.19 0.48 0.42 0.38 0.48

Industry year C15 C16 C17 C18 C19 C20 C21 Average

1.50 1.03 0.41 0.28 0.73 0.58 0.77

Industry year C22 C23 C24 C25 C26 Average

0.75 1.39 1.20 1.62 1.46

Note: Corresponding of codes and categories: C1 (food processing and manufacturing industry), C2 (beverage manufacturing industry), C3 (tobacco industry), C4 (textile industry), C5 (textile and garment, shoes, hats manufacturing industry), C6 (leather fur and feather industry), C7 (timber processing and furniture manufacturing industry), C8 (papermaking and paper products industry), C9 (printing and record medium reproduction), C10 (cultural educational and sports goods), C11 (petroleum refining and coking), C12 (chemical raw materials and chemical products manufacturing industry), C13 (pharmaceutical manufacturing industry), C14 (chemical fiber industry), C15 (rubber product industry), C16 (plastic product industry), C17 (non-metallic mineral products), C18 (black metal smelting and rolling processing industry), C19 (non-ferrous metal smelting and rolling processing industry), C20 (fabricated metal products), C21 (ordinary machinery manufacturing industry), C22 (special equipment manufacturing), C23 (transportation equipment manufacturing industry), C24 (electrical machinery and equipment manufacturing industry), C25 (communication equipment computers and other electronic equipment manufacturing industry), C26 (instrumentation and cultural office machinery). Source: Calculated based on IIR network database data.

DESCRIPTION OF THE DATA SOURCE AND INDUSTRY SELECTION

3.1 Industry selection According to the National industrial classification, we select the manufacturing industries except arts and crafts industry and the processing of waste recycling

50

3.2

Table  3. Tendency of international fragment of China’s manufacturing production based on the measurement of processing trade import and export from 2001 to 2008.

Data source

We can obtain the data in equation (1) from the IIR network database such as import and export of China’s processing trade imports and exports and manufacturing data. But IIR network database data on trade are classified by HS codes (Harmonization Code System Code), and are not according to national industrial classification. Referring to Cheng Bin (2002), we took the related merchandise trade data in correspondence with the 26 selected manufacturing industries. Then we obtained the processing trade data of the industries. Processing trade includes processing and assembling trade, processing trade, processing trade and imports of foreign-invested enterprises as an investment in equipment and supplies. The time span of China HS code of 4-digit data available in IIR network is from 2001 to 2008. 4

Industry year C1 Average

Average

Average

Average

C5

C6

C9

C7

C10 C11 C12 C13 C14

0.69 0.61 1.44 0.05 0.72 0.04 1.49

Industry year C15 C16 C17 C18 C19 C20 C21 Average

0.65 1.12 1.33 0.69 0.91 1.03 0.80

Industry year C22 C23 C24 C25 C26 Average

C7

C9

C10 C11 C12 C13 C14

0.75 1.15 1.25 0.24 0.60 0.18 0.89

1.28 1.07 0.61 0.45 0.83 0.68 0.78

0.81 0.85 1.24 1.47 1.37

tries’ trends above 1  such as C8, C9, C10, C15, C16, C23, C24 and C26. Other industries’ trends are relatively lower, such as C2, C3, C17, C18, C12 and C13. Table 2 shows the tendency of international fragment of China’s manufacturing production from 2001 to 2008 measured by the import of processing trade. Among them, C6 has the highest trend. There are some industries’ trends above 1 such as C4, C5, C10, C14, C16, C17, C24, C25 and C26. Other industries’ trends are relatively lower, such as C2, C3, C11, C13 and C23. Table  3  shows the tendency of international fragment of China’s manufacturing production from 2001 to 2008 measured by the total amount of import and export of processing trade. Among them, C25 has the highest trend. In addition, there are some industries’ trends above 1  such as C9, C10, C15, C16, C24 and C26. Other industries’ trends are relatively lower, such as C2, C3, C11 and C13. Obviously international fragment of production measured by the total amount of import and export of processing trade reflects the general participation of China’s manufacturing industries. But there are different tendencies of international fragment of China’s manufacturing production measured by the import or export of processing trade. It also shows that some industries participate in the international fragment of production mainly through the processing trade exports, such as C8, C9, C15, and C23. While others participate in the international fragment of production mainly through the processing trade imports, such as C6, C4, C5, C14 and C17. Other participations of imports and exports are more prominent, such as C10, C16, C24, C25 and C26.

0.43 0.03 0.28 1.94 1.32 1.71 0.91

Industry year C8 Average

C4

C6

Source: Calculated based on IIR network database data.

Table 2. Tendency of international fragment of China’s manufacturing production based on the measurement of processing trade import from 2001 to 2008. C3

C5

Industry year C22 C23 C24 C25 C26

In equation (1), we use the 26  manufacturing industries’ data such as export, import, import and export trade, processing trade import and export, and total amount of import and export to calculate and measure the degree of international fragment of China’s manufacturing production from the point of import and export (see Tables 1–3). Table 1 shows the tendency of international fragment of China’s manufacturing production from 2001 to 2008 measured by the export of processing trade. From the perspective of average, C25 has the highest trend. In addition, there are some indus-

C2

C4

Industry year C15 C16 C17 C18 C19 C20 C21

DEGREE OF INTERNATIONAL FRAGMENT OF CHINA’S MANUFACTURING PRODUCTION BASED ON PROCESSING TRADE

Industry year C1

C3

0.50 0.12 0.09 0.70 0.72 0.88 0.73

Industry year C8 Average

C2

0.85 0.09 1.30 1.27 1.36

Source: Calculated based on IIR network database data.

51

5

CONCLUSIONS

degree of participation in the international fragment of production, and international fragment of production greatly accelerated the growth of manufacturing output and exports in China. Meanwhile, it also showed that China’s manufacturing output and exports contained large amount of the foreign investment. After the deduction of foreign intermediate inputs, the real added value and export of Chinese manufacturing industry might not be so much, and so does the trade interests. So, what kind of stage should Chinese manufacturing production sections are in the international division of production system, and how to strive for a higher degree of fragmentation in the production with high added value and more interest? This research direction is worthy of further investigation.

The advantages of using metrics based on processing trade data to measure the degree of China’s participation in the international fragmentation of production is availability of data and that processing trade imports and exports can comprehensively reflect the situation and the characteristics of China’s manufacturing industries’ participation in the international fragment of production. Measurements based on processing trade data shows that from the perspective of overall imports and exports, seven manufacturing industries have higher degree of international fragmentation of production, respectively: communication equipment, computers and other electronic equipment manufacturing industry, printing and record medium reproduction, cultural and educational sporting goods manufacturing industry, the rubber products industry, the plastic products industry, electrical machinery and equipment manufacturing industry, instrumentation and cultural office machinery. Other four industries have lower degree of international fragmentation of production, respectively: beverage manufacturing industry, tobacco industry, oil processing and coking industry and pharmaceutical industry. Papermaking and paper products industry, printing and record medium reproduction, rubber Product industry and transportation equipment manufacturing industry participated in the international fragmentation of production through the export of processing trade. Industries of leather fur and feather, textile industry, textile and garment, shoes, hats manufacturing industry, chemical fiber industry and non-metallic mineral products participated in the international fragmentation of production through the import. Cultural educational and sports goods, plastic product industry, electrical machinery and equipment manufacturing industry, communication equipment, computers and other electronic equipment manufacturing industry and instrumentation and cultural office machinery participated in the international fragmentation of production through both the import and export. Obviously, most of China’s manufacturing industry had relatively high

ACKNOWLEDGMENT This research is supported by Natural Science Foundation of China (No. 71163047). REFERENCES Buckley, P.J. The Impact of the Global Factory on Economic Development [J]. Journal of World Business, 2009, (44):131–143. Egger, H. & Egger, P. (2003), “Outsourcing and skillspecific employment in a small economy: Austria after the fall of the Iron Curtain”, Oxford Economic Papers 55(4), 625–643. Feenstra, R.C. & Hanson, G.H. (1996), “Globalization, Outsourcing, and Wage Inequality,” American Economic Review, 86(2), 240–245. Hummels, D., Rapoport, D. & Yi, K.-M. (1998), “Vertical specialization and the changing nature of world trade”, Federal Reserve Bank of New York Economic Policy Review 4(2), 79–99. Hummels, D., Ishii, J. & Yi, K.-M. (2001), “The nature and growth of vertical specialization in world trade”, Journal of International Economics 54(1), 75–96. Maskell, P., Pedersen, T., Petersen, B. & Dick-Nielsen, J. Learning Paths to Offshore Outsourcing: from Cost Reduction to Knowledge Seeking [J]. Industry and Innovation, 2007,14, 14(3):239–257.

52

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

The voltage-controlled low-pass filter based on FPGA frequency measurement Sheng-Qian Ma, Li-Rong Zheng, Juan-Fang Liu & Yan-Ping Ji Key Laboratory of Atomic and Molecular Physics and Functional Materials of Gansu Province, College of Physics and Electronic Engineering, Northwest Normal University, Lanzhou, Gansu, China

ABSTRACT: This paper describes the circuit and implementation method of the voltage-controlled self-tracking low-pass filter, which uses FPGA to measure the frequency of the input signal, and uses a special method to make frequency convert to corresponding voltage by D/A converter. Then the voltage signal is input into voltage-controlled low-pass bi-quad loop filter circuit which is mainly composed of an analog multiplier AD835 and the current feedback operational amplifier OPA658. Cut-off frequency of the filter can automatically track the input signal frequency changes by adjusting the external control voltage through FPGA programming. This paper describes the system work principles, infers the transfer function of the filter and designs the circuit of voltage-controlled second-order low-pass filter. The filter can realize the scope of the cut-off frequency up to 8 MHz. The simulation and experiment results show the valid of the filter system designed. Keywords: FPGA frequency measurement; D/A; frequency automatically tracking; voltage-controlled low-pass filter 1

INTRODUCTION

difficult to achieve the tracking of the frequency of the input signal. This paper proposes a voltage-controlled selftracking filter system that can control the cut-off frequency of the filter by changing the external voltage. In the system, FPGA is used to measure frequency of the input signal, and D/A convertor is used to convert the measured frequency to voltage signal which will be input into the bi-quad loop filter as the external control terminal to control the cut-off frequency to track the input signal frequency changes automatically. Compared the voltage-controlled filter designed in this paper with other voltage-controlled filters, its cut-off frequency can dynamically adjust, and filtering speed is high, waveform is smooth and other features.

The filter has a wide range of applications in the fields of the signal processing, the data collection and the communication systems. Especially, the self-tracking filter plays an important role in the field of the adaptive signal processing. There are many kinds of methods to realize the automatic self-tracking filter. The first is the state variable method which can be carried out simultaneously low-pass, high-pass and band-pass filter, but the design and debug are troublesome and difficult to control, and besides the components are discrete and the non-linear effects are large. The second method is the voltage-controlled method. Using this method when the input signal frequency is within the bandwidth of the filter, there is no phase difference between the filtered signal and the input signal. But the offset voltage and drift will directly affect the stability of the low frequency. The third method is using existing active filter chip. This method does not require to know the signal frequency in advance, and filtering frequency is controllable, but there are some circuit noise and signal aliasing problems in the course of the use, even through clock frequency or pin programming is difficult to achieve the filter cut-off frequency continuous change. If the signal frequency changes in a wide range, the above three methods are

2

FREQUENCY VOLTAGE CONVERSION

The frequency voltage conversion circuit is composed of FPGA and D/A. FPGA measures the frequency of input signal, and then D/A converts the measured frequency into voltage. It is necessary to establish a linear relationship between the measured frequency and the converted voltage. Then we can obtain the voltage corresponding to different frequency by finding the scale factor.

53

2.1

Frequency measurement circuit based on FPGA

The frequency measurement circuit uses the EP2C20F484C8  N of Altera Corporation as the core chip. The chip have 50 MHz active crystal and 50 MHz active clock inside. The measured signal fin is amplified, shaped and limited by signal preprocessing module, and the shaped rectangular pulse fx is input into FPGA as the input signal. The diagram of FPGA frequency measurement is shown in Figure 1. The crystal oscillator provides 50  MHz clock signal fs. The signal fs is divided into two paths, the first inputs into counter 2 as clock pulse, the other one makes frequency division through signal source module and generates 1  second gate time Td, which is regarded as the clock input of the control module. The control module generates the count enable signal EN and the reset signal CLR. When measuring, the first thing is resetting the two counters. When the signal EN is for high level and the rising edge of fx signal to be measured is arrived, counters start to count the measured signal and the clock signal separately. When EN is for low level and the rising edge of fx signal to be measured is arrived, counters stop counting. The count value is latched and sent to display unit. Assuming within a gate time Td, the count value of the measured signal counted by the counter is Nx, and the count value of the clock signal is Ns, then the frequency N x /N s fs . of the measured signal is fx 2.2

Figure  2. The relation of linear frequency-to-voltage convert.

Table 1. The measured and theoretical voltage values of the frequency voltage conversion circuit. fx/Hz

2M

4M

6M

8M

Vx/V Theoretical Measured

0.250 0.248

0.500 0.479

0.750 0.748

1.000 0.996

Relative error δ (%) Absolute error/V

0.800 0.002

0.600 0.003

0.267 0.002

0.400 0.004

in order to achieve a good linear relationship, we design that when the measured frequency is 8 MHz, the converted voltage is 1.00 V, and when the measured frequency is 0  Hz, the converted voltage is 0.00V, thus we can get that the conversion relationship between the measured frequency and the converted voltage is

D/A conversion circuit

D/A conversion circuit is mainly responsible for converting the digital frequency measured by FPGA to the corresponding analog voltage, and loading the voltage signal into the control terminal of the voltage-controlled filter, it can change the cut-off frequency of the filter, so as to achieve the goal of the cut-off frequency automatically tracking the input signal frequency. This design uses a programmable dual 12 bit converter, that is TLC5618 AC D/A converter. Supposing the measured frequency fx is from 0 Hz to 8 MHz, the voltage is from 0.00 V to 1.00 V,

fx = 8 × 106 Vx

(1)

From formula (1), we can calculate the corresponding conversion voltage. The relation of linear frequency-to-voltage convert is shown in Figure 2. The measured and the theoretical voltage values of the frequency voltage conversion circuit is shown in Table 1. From the table, we can know the error of the frequency voltage conversion circuit is small. 3

DESIGN OF VOLTAGE-CONTROLLED TRACKING FILTER

The system diagram of the voltage-controlled tracking filter is shown in Figure 3. The input signal fin is divided into two paths, the first is transformed into rectangular wave signal fx through amplifying, amplitude-limiting and shaping, and put fx into FPGA to measure the frequency, then use the special linear frequency-to-voltage conversion method to control D/A convert the measured

Figure 1. The diagram of FPGA frequency measurement.

54

U4, R5 and R7 can reduce the closed-loop gain, improve the stability of gain and decrease the nonlinear distortion. U1 can achieve the square of the signal, which makes a linear relationship between the cut-off frequency and the input frequency. U1 and U3 are both four quadrant analog multiplier AD835, U2, U4 and U5 are low power current feedback operational amplifier OPA658, D0, D1, D2 and D3 are resistance network, Vx is voltage control terminal, fin is input signal terminal, fout is filtered signal terminal. According to Kirchhoff’s law, we can obtain: Figure 3. The system diagram of the voltage-controlled tracking filter.

VW = Vx2

(2)

Vout/R5 = VG/R7

(3)

VWVF = VE = VoutC1sD2

(4)

Vin/D0 = VF/D3 + VFC2s + VG/D1

(5)

According to the formula (2–5), the transfer function of the filter is:

H (s) =

Vout Vin

Vx2 C1C2 D0 D2 = Vx2R7 s 2 s + + C2 D3 C1C2 D1D2R5

(6)

According to the formula (6), it can deduce the cut-off angular frequency is ωLP = (VxR71/2)/(C1C2D1D2R5)1/2 Figure  4. The diagram of the voltage-controlled second-order low-pass filter circuit.

the gain factor is K = (R7D0)/(R5D1)

frequency into the corresponding analog voltage Vx, which is the voltage control terminal of the filter. The other one is as the input signal of voltage-controlled filter, so that the combination of the two signals can realize the function of the frequency tracking automatically.

(8)

the quality factor is QLP = C2D3ωLP

(9)

so the cut-off frequency is fLP = (VxR71/2)/[2π (C1C2D1D2R5)1/2]

3.1

(7)

Design of voltage-controlled second-order low-pass filter

(10)

When taking C1  =  C2  =  C, D0  =  D1  =  D2  =  D, R5 = R7, it can deduce

The second-order voltage-controlled filter circuit is achieved by using the analog multiplier and introducing into bi-quad loop filter circuit model, its circuit is shown in Figure  4. The addition of damping integrator composed by D0, D1, D3, C2 and U2 plays a role of automatic compensation. The inverting integrator composed by D2, C1 and U5 is used as the linear modulator to complete the pulse width demodulation. The inverting input closed-loop operational amplifier composed by

fLP = Vx/2πDC

(11)

The formula (1) is substituted into formula (11), it can obtain fLP = fx/(8 × 106 × 2πDC)

(12)

It can be seen from (12) the cut-off frequency of the filter designed will change linearly with input

55

Figure 5. The experimental and theoretical values comparison of the amplitude-frequency responses of the voltagecontrolled second-order low-pass filter.

frequency as long as adjusting the value of capacitance and resistance reasonably. 3.2

Table 2. The comparison of the measured and theoretical values of the cut-off frequency of the second-order low-pass filter. f0/KHz

The experimental result

In the actual measurement, when taking C1 = C2 = C = 100pf, D0 = D1 = D2 = D = 200Ω, D3 = 115Ω, R5  =  R7  =  500Ω, then according to formula (12), it can deduce fLP  =  fx, fx is the frequency signal after fin amplifying, amplitude-limiting and shaping, so fLP  =  fx  =  fin, that is, it realizes the cut-off frequency to track the input frequency. From formula (11), it can deduce the cut-off frequency of filter is fLP = 7.9577 × 106Vx. The amplitude of the input signal is taken 1.0 V, the control voltage Vx are taken 0.1 V, 0.4V, 0.7 V, 1.0 V, for physical testing, the actual measured corresponding cut-off frequency are 796.2 KHz, 3184.7 KHz, 5573.3 KHz, 7961.8  KHz, the experimental and theoretical values comparison of the amplitude-frequency responses is shown in Figure 5. From Figure  5, it can be seen that when the second-order voltage-controlled low-pass filter taking different control voltage Vx, the cut-off frequency will realize to change and track the input signal. The advantages of this method are wide range of adjusting and tracking and high precision. Detailed error analysis is shown in Table 2.

Vx/V

Measured

Calculated

Relative error δ (%)

0.1 0.4 0.7 1.0

796.20 3184.70 5573.30 7961.80

795.77 3183.08 5570.39 7957.70

0.054 0.051 0.052 0.051

4

CONCLUSION

This paper describes a method that firstly uses FPGA to measure the frequency of input signal, and uses the special method to control D/A to convert the measured frequency into the corresponding voltage, then inputs this voltage signal into analog multiplier which is a filter unit to control the cut-off frequency of the filter indirectly. Based on the proposed design method, we design self-tracking second-order low-pass filter circuit, the input signal frequency ranges up to 8 MHz to ensure the cut-off frequency of the filter can track automatically the input signal. The experimental result matches well with the theoretical value,

56

which can prove the correctness and effective of circuit design. The structure of circuit is simple and conveniently controlled. And the filter system has an ability of fast tracking input signal change, and good real-time dynamic characteristics. It can be widely used in all aspects of the signal processing.

Jiang. Z.P. 2010. Design and implementation of FM exciter. Communication technology 9: 50–53. Li. X.P, Xu. J, Li. J. 2006. The principle and implementation method of the voltage-controlled filter. Modern Manufacturing Engineering 10: 114–115. Ma. S.Q, Ran. X.P, Fan. M.H. 2013. The design of the self-adaptive low-pass filter. Piezoelectrics & Acoustooptics 35(2): 245–249. Rezaei. F, Azhari. S.J. 2011. Ultra low voltage, high performance operational transconductance ampliier and its application in a tunable Gm-C ilter. Microelectronics Journal 42: 827–836. Shi. X.C, Liu. S.L, Yu. M. 2010. The design of the antialiasing filter using the monolithic integrated active filter chip. Automation and instrumentation 3: 10–11. Tao. L.M, Li. Y, Wen. X.S. 2007. Signal tracking filter method based on technology of switched capacitor and its application in Rotor Balancing. China Mechanical Engineering 18(4): 427–430. Yang. Z.M. 1999. A basic method of the circuit based on the operational amplifier converted into the circuit based on CFA. Journal of Northwest normal university (natural science edition) 35(2): 31–38. Yu. W.W, Yan. D.T, Yang. Y. 2008. The engineering design and application of the automatic tracking filter based on MAX260. Modern electronic technology 31(5): 76–78.

ACKNOWLEDGEMENT In this paper, the research work was sponsored by the Natural Science Foundation of China (61162017), Department of Education-fund projects in Gansu (1101–03) and Northwest Normal University NWNU-LKQN-13-16. REFERENCES Guo. C.P, Ni. W.Q. 2013. The stability analysis of voltage-controlled voltage source filter. Journal of Luoyang Normal University 32(5): 32–34. Horng. J.W, Hou. C.L. 2005. Voltage-mode universal biquadratic ilters with one input and ive outputs using MOCCIIs. Computers and Electrical Engineering 31: 190–202.

57

This page intentionally left blank

Mechatronics Engineering and Electrical Engineering – Sheng (Ed.) © 2015 Taylor & Francis Group, London, ISBN 978-1-138-02719-0

Departure capacity assessment of close staggered parallel runways J.G. Kong, X. Li & W.B. Ding School of Air Traffic Management, Civil Aviation Flight University of China, Guanghan, Sichuan, China

ABSTRACT: First, this article analyzes the staggered layout of closely spaced parallel staggered runways, its advantages compared to the traditional closely spaced parallel runway system and operating mode of the departure of aircraft. According to its operational characteristics, this paper learns from the departure capacity calculation model under mixed operation mode of single runway to establish the departure capacity calculation model of closely spaced parallel staggered runways. Then, it takes a group of closely spaced parallel staggered runways of Shanghai Pudong airport which consists of runway 1 and runway 3 as example, collects the relevant operational data, and uses the established model to do the data simulation to obtain the departure capacity of this parallel runway system. Keywords: 1

closely spaced parallel runway; staggered; departure capacity one runway for departure, the other runway for arrival, to establish a correlation between departure and arrival safety interval. The implementation of this way of running parallel runways called closely spaced parallel runway, referred to closely runway. In a certain mode of operation, flight procedures, air traffic control and communications and navigation equipment conditions, closely spaced parallel runway capacity is much larger than a single runway, but the land used for significantly less than the distance parallel runways. Accordance with the provisions of , closely runway’s operation mode only with one arrival one departure, and Article 43 of