摘要
随着我国城市化进程的不断加快,城市规模越来越大,城市人口不断增多,现有的城市交通网络已经难以满足城市发展的要求。根据国际上大型城市交通发展的经验,轨道交通对于缓解城市交通压力,增加城市交通活力起着举足轻重的作用。在此背景下,因为列车自动驾驶(ATO)系统具有提高轨道交通运行效率,加快列车运行速度,保证行车安全等功能,开发高效智能的列车自动驾驶系统就显得至关重要。
本文讨论了列车自动驾驶系统的基本结构、功能和列车自动驾驶的操纵原则,指出应用基于传统控制理论的PID控制器来进行已不能适应列车运行参数的非线性和时变性,很难真实反映实际情况,从而使速度控制的平滑性受到破坏,牵引和制动的切换频繁,影响了系统舒适性的要求,同时也影响了列车的节能性能和停车精度。而采用先进的智能控制方法对传统的控制器及其追踪曲线进行优化,则可有效解决上述问题。
模糊控制、神经网络、遗传算法是近些年来发展较快的三种智能算法,能在不同程度上模仿司机的操作行为,因此可以用来优化和改进传统的ATO控制系统。本文依次介绍了模糊控制、神经网络和遗传算法在优化PID控制系统及其目标曲线中的应用,并用Matlab和Simlink工具箱进行了建模仿真。主要研究内容如下:
(1)通过对列车自动驾驶系统结构和功能的研究,总结列车自动驾驶系统的驾驶策略及优化操纵原则,分析ATO系统的所要达到的性能指标,在此基础上对列车运行进行建模。
(2)设计列车自动驾驶算法。首先根据传统的控制算法设计PID控制器,跟踪列车目标曲线行车,在此基础上引入对传统PID控制器及其目标曲线进行优化和改进的智能控制算法:模糊控制算法、神经网络控制算法和遗传算法控制算法。
(3)依次实现三种智能算法在ATO系统中应用的仿真建模。根据ATO系统优化操纵所要达到的五项性能要求,依次基于以上三种智能算法,为ATO系统设计更优秀的PID控制器和更合理的目标曲线,并对其性能进行了评价分析。
(4)对优化了的ATO系统的整体控制效果进行仿真验证。在选取实际线路参数和列车参数的情况下,基于上述三种智能算法仿真所得到的结果,分析优化后的ATO系统控制的列车运行曲线,得出智能算法合理性的结论。
As China's urbanization process, the continuing development of city size and increase of the urban population, existing urban traffic net is unable to meet the demands of city traffic. According to the international large-scale urban transportation developing experience, UMT plays the pivotal role for easing urban traffic pressure. Under this background, it is important to develop the the highly effective intelligence ATO system because ATO can improve the operationg efficiency, speed up the train pace, and ensure the safety.
This article discussed the ATO system's basic structure, function and the automatic operation principle of the train to point out that using the PID controller based on the tradition control theory has cannot adapt the train movement parameters' misalignment and time-variable. It is very difficult to reflect the actual situation really, thus causes the speed control smoothness to receive the destruction, the hauling and the brake cut is frequent, and affects the system comfortable request. Simultaneously has also affected train's energy conservation performance and the parking precision. But we can solve the above questions effectively by using the advanced intelligent control method to carries the optimization on the traditional controller and the tracing curve.
The Fuzzy Control, the Neural Network and the Genetic Algorithm are the three kind of intelligent algorithms which develop quickly in recent years. They can imitate driver's operation behavior in varying degrees, therefore can be used to optimize and improve the traditional ATO control system. This article introduced the Fuzzy Control, the Neural Network and the Genetic Algorithm's using in optimizing the PID control system and the target curve in turn, and has carried on the modelling simulation with Matlab and the Simlink toolboxs. Main contents are as follows:
(1) Through the research of the ATO system's structure and the function, summarize the ATO system's driving strategy and the optimized operation principle, analyze performance index which the ATO system must achieves, and on this basis we carry on the modelling to the train movement.
(2) Design the ATO algorithm. First we design the PID controller according to the traditional control algorithm so as to track the train goal curve driving, on this basis introduce the intelligent control algorithms: the Fuzzy Control, the Neural Network and the Genetic Algorithm, which are used to the optimize and improve the traditional PID controller and the target curve.
(3) Realizes the three kind of intelligent algorithms' application in the ATO system's simulation modelling in turn. On the basis of the above three kind of intelligent algorithms, design the more outstanding PID controller and the more reasonable target curve according to five performance requirements which the ATO system optimization operation must achieve, and carriy on the appraisal analysis to its performance.
(4) Test the optimized ATO system's whole control effection by simulation and confirmation. In the situation of selection actual line parameters and train parameters, analyze the ATO systems control train movement curve which has been optimized based on the above three kind of intelligent algorithm's simulation, after that draw the conclusion of the intelligent algorithm's rational.
引文
[1]郜春海,唐涛,张建明.高速铁路列车运行控制系统车载设备的软件设计[J].北京交通大学学报,1999(5).1-2
[2]宗明,郜春海,何燕.基于CBTC控制的全自动驾驶系统[J].都市快轨交通,2006(6).1-3
[3]唐涛,黄良骥.列车自动驾驶系统控制算法综述[J].铁道学报,2003(4).1-5
(4]康太平,王长林.基于模糊预测控制的列车自动驾驶系统研究[D].成都:西南交通大学,2006(3).33-46
[5]石红国,彭其渊.列车运行过程仿真及优化研究[D].成都:西南交通大学,2006(9).9-27
[6]吴玉生,龙志强.磁浮列车自动驾驶(ATO)系统控制算法研究与仿真[D].湖南:国防科学技术大学,2006(4).18-29
[7]何庆,冯晓云.基于遗传算法和模糊专家系统的列车优化控制[D].成都:西南交通大学,2006(4).8-19
[8]韦巍.智能控制技术[M].北京:机械工业出版社,2000(1).112-124
[9]黄良骥,程琳香,唐涛.遗传算法模糊神经网络在列车驾驶中的应用[J].辽宁工程技术大学学报,2001(10).1-4
[10]孙增圻.智能控制理论与技术[M].北京:清华大学出版社,1992(4).345-373
[11]C.S.Chang,S.S.Sim.Optimising train movements through coast control using genetic algorithms[J].IEE Proc.-Electr.Power,1997(1).1-9
[12]崔世文,冯晓云.列车优化操纵与自动驾驶模式的研究与仿真[J].铁道机车车辆,2005(10).1-4
[13]杨光,唐祯敏.基于MATLAB的磁浮列车自动驾驶控制系统的仿真[J].轨道交通信息系统.2007.1-3
[14]黄良骥,唐涛.地铁列车自动驾驶系统分析与设计[J].北京交通大学学报,2002(6).1-4
[15]朱波,高振海.基于驾驶员行为的车道保持系统神经网络决策与控制[D].吉林:吉林大学,2006(4).51-71
[16]刘亚营,刘以建.改进型遗传算法及其在神经网络参数优化中的应用[D].上海:上海海运学院,2003(12).23-36
[17]王洪坡,尹力明.磁悬浮列车速度控制与自动驾驶系统研究[D].湖南:国防科技大学,2002(11).69-74
[18]陈霞,魏忠义.网络化通用自动控制系统设计[J].西安:西安工程科技学院,2006(5).1-4
[19]丘焕耀,毛宗源.模糊控制、神经网络和变结构控制的交叉结合及其应用研究[D].广东:华南理工大学,1999(5).1-13
[20]张邦楚,韩子鹏.基于分数阶微积分的飞航式导弹控制系统设计方法研究[D].南京:南京理工大学,2005(10).85-112
[21]何海,钟毅芳.混合动力汽车控制系统设计与仿真[D].湖北:华中科技大学,2005(9).82-99
[22]刘金琨.先进PID控制MATLAB仿真[M].北京:电子工业出版社,2004(9).210-247
[23]许力.智能控制与智能系统[M].北京:机械工业出版社,2006(12).296-333
[24]Tang Tao,Gao Chunhai.A New Train Speed Automatic Control System For Chinese Railway[J].Proceedings of The IEEE International Conference on Industral Technology,1996
[25]贾利民,张建华,张锡第.高速铁路列车运行控制的现状与展望[J].中国铁道科学,1996(12).95-100
[26]板仓荣治.日本铁道ATC技术发展概述[J].中国铁路,1997(12).48-52
[27]渡边郁夫.日本铁道数字ATC系统的开发[J].中国铁路,1998(1).46-49
[28]吴汶麒.城市轨道交通信号与通信系统[M].北京:中国铁道出版社,1998
[29]黄玲珍.列车运行仿真软件的研制及其操纵优化的探讨[D].上海:同济大学,1998.10-25
[30]Zadeh.On Optimal Control and Linear Programming.IRE Trans.Aut.Contril,1962(7).45-49
[31]朱立海.平稳操纵机车及其控制装置[J].铁道运营技术,1996(2).160-162
[32]刘毅.对列车启动时平稳操纵的讨论[J].内燃机车,2003(8).36-37
[33]刘利.使用电阻制动是确保列车实现安全与平稳操纵的根本途径[J].内燃机车,1998(12).13-15
[34]张波.中高速列车共线运行的仿真研究[D].北京:铁道科学研究院,2000.15-22
[35]中华人民共和国铁道部.机车操作规程.北京:中国铁道出版社
[36]徐安.城市轨道交通电力牵引[M].北京:中国铁道出版社,2000.50-52
[37]练级三.电力牵引控制系统[M].北京:中国铁道出版社,1994.20-32
[38]克里斯坦森,塞耶 软件工程最贱实践项目经理指南[M].北京:电子工业版社,2004.30-62
[39]内田清五.日本新干线列车制动系统[M].北京:中国铁道出版社,2004.20-22
[40]Yoshio YUYAMA.The Tobu Kyuryo Line A magnetic Levitaion system.Maglev's Presentaions,2004.308-315
[41]张威.Matlab基础与编程入门[M].西安:电子科技大学出版社,2004.10-52
[42]邓聚龙.灰色控制系统[M].湖北:华中理工大学出版社,1997.10-13
[43]王伟.广义预测控制及应用[M].北京:科学出版社,1998.1-62
[44]王耀南,宋明,童调生.一种基于神经网络的自组织模糊控制及应用[J].模式识别与人工智能,1994(7).285-292
[45]杨国勋,郭晨,贾欣乐.基于增强型学习算法的船舶运动混合智能控制[J].中国航海,2001.321-324
[46]马良,蒋馥.多目标旅行售货员问题的蚂蚁算法求解[J].系统工程理论方法应用,1999(8).23-26
[47]谭浩强.C程序设计[M].北京:清华大学出版社,1998.1-200
[48]徐士良.计算机常用算法[M].北京:清华大学出版社,1995.10-50
[49]龚剑,朱亮.Matlab5.X入门与提高[M].北京:清华大学出版社,2000.1-120
[50]Satoshi Sekine,Naoki Imasaki,Tsunekazu Endo.Application of Fuzzy Neural Network Controlto Automatic Train Operation and Tuning of Its Control Rules.IEEE,2005.1-2
[51]李博轩.Visual C++图形用户界面开发指南[M].北京:清华大学出版社,2000.1-99