摘要
对于交通流监控和诱导系统来说,交通流状态辨识和交通流短时预测是其中最为关键的组成部分之一。这些子系统性能的好坏,直接关系到交通流监控及诱导系统的准确性和有效性。因此,在ITS技术的基础上,对这两个部分进行研究,具有十分重要的现实意义。
文章首先分析了交通流的主要特征参数,并研究了各种参数的监控方法,并在此基础上,讨论了交通流信息监测中存在的几个难点和解决方法；之后,在对交通流状态辨识和交通流短时预测技术进行分析的基础上,详细研究了基于支持向量机的交通流状态辨识技术和交通流短时预测技术,对于交通流状态辨识的系统框架、数据处理方法、交通流短期预测的系统框架、数据处理方法都进行了研究,并引用实例进行了验证分析。文章研究发现支持向量机在小样本容量的分类问题上表现良好,泛化能力较强,在合理的参数下精度较高；而在回归的应用中视样本数据的质量,精度有波动,但是依然在一定精度下可以完成数据的拟合工作。
研究结果表明,支持向量机在交通流状态辨识技术和交通流短期预测技术中存在现实的应用价值,并且其准确性、精度和推广能力都达到了一定的要求,结果是可信且有效的。应用支持向量机技术的交通流管控系统不仅在理论上可行,现实中也可推广应用。
For traffic flow monitoring and guidance system, traffic flow state identification and traffic flow forecasting are one of the most critical component of it. The quality of the performance of these subsystems is directly related to the accuracy and effectiveness of it. Therefore, based on the ITS technology, the study of these two parts is of great practical significance.
The article firstly analyzes the main features of traffic flow parameters, and the method of monitoring of various parameters. On this basis, several difficulties and the solutions in monitoring of traffic flow information are discussed. Later on, based on the analysis of the identification of the state of traffic flow and traffic flow forecasting techniques, a detailed study of support vector machine based traffic state recognition technology and traffic flow forecasting technology has been done. The traffic flow system framework of state identification, data processing, traffic flow short-term forecasting system framework, data processing methods were studied and cites examples to validate analysis. Article found that support vector machine in the classification of the small sample size performs well and has high precision in reasonable parameters. While the application of the regression, as the sample data quality, accuracy changes, but the work still can be done under a certain accuracy of data fitting.
The results show that support vector machine has actual value in the recognition technology in the state of traffic flow and traffic flow short term forecasting techniques. Its accuracy, precision and generalization ability meet the certain requirements, the results are credible and effective. Application of support vector machine control system of traffic flow is not only theoretically possible, but also can be generalized in reality.
引文
[1]刘静.基于状态划分的交通流短时预测方法研究[D].北京：北京交通大学交通运输学院，2006
[2]王凡.基于支持向量机的交通流预测方法研究[D].大连：大连理工大学,2010
[3]刘静,关伟.交通流预测方法综述.公路交通科技,Vol.21No.32004年3月
[4]Brian L.Smith, Michael J.Demetsky. Traffic Flow Forecasting:Comparison of Modeling Approaches[J]. Journal of Transportation Engineering,1997,123(4):261-266.
[5]Howard R. Kirby, Susan M.Watson and Mark S.Dougherty. Should we use neural networks or statistical models for short-term motorway traffic forecasting[J].International Journal of Forecasting, 1997,13(1):43-50.
[6]Bart Van Arem, Howard R.Kirby, Martie J.M.Van Der Vlist, Joe C.Whittaker. Recent advances and application in the field of short-term traffic forecasting[J]. International Journal of Forecasting,1997,13(1):1—12.
[7]Lingras Pawan, Sharma Satish C., Subramanian Gopalakrishnan. Short-term traffic volume forecasts:existing and future research[C]. Canadian Society for Civil Engineering—1999 Annual Conference. Montreal, Canada:Canadian Society for Civil EnQineering,1999:429-438.
[8]Lee S, Cho B, Hong S. Dynamic adaptive short-term vehicle speeds prediction models[C].3rd Conference of the Eastern Asia Society for Transportation Studies. Taipei,Taiwan,1999. Tokyo, Japan: Eastern Asia Society Transportation Studies.1999:273-287.
[9]Sherif Ishak, Haitham AI-Deek. Performance Evaluation of Short-Term Time-Series Traffic Prediction Model [J]. Journal of Transportation Engineering,2002,128(6):490-498.
[10]重庆交通学院.交通流主要参数及交通事件动态预测算法研究报告[R].重庆：重庆交通学院,2005
[11]李硕.高速公路常发性交通拥挤路段实时判定与跟踪研究[J].中南公路工程,2005,3(30)：158-161
[12]黄中祥,王正武,况爱武.短时交通流可预测性分析与比较[J].土木工程学报,2004,37(2)：108-113
[13]杨兆升.城市交通流诱导系统理论与模型[M].北京：人民交通出版社,2000
[14]杨忠振,孙卓.基于GIS的城市道路交通管理信息系统[J].交通与计算机,2002,20(5)：18—21
[15]姜桂艳.道路交通状态判别技术与应用[M].北京：人民交通出版社,2004
[16]贺国光,李宇,马寿峰.基于数学模型的短时交通流预测方法探讨[J].系统工程理论与实践,2000,12：51-56
[17]邓乃扬,田英杰.数据挖掘中的新方法-支持向量机[M].北京：科学出版社,2004.
[18]黄发良,钟智.用于分类的支持向量机[J].广西师范学院学报(自然科学版),2004,21(3)：75-78.
[19]夏冰,董著,张佐.周相似特性下的交通流预测模型研究[J].公路交通科技,2003,(2)：73—76
[20]籍东辉.城市道路无检测器交叉路口交通流量预测方法研究[D].长春：吉林大学,2002
[21]安实,王健.城市道路动态信息采集系统结构设计[J].交通运输系统工程与信息,2002,Vol.2No.02
[22]于春全.北京市道路交通流实时动态信息系统的研究[J].交通运输系统工程与信息,2002,Vol.2No.3
[23]姜桂艳.动态交通数据故障识别与修复方法[J].交通运输工程学报,2004,4(1:)121—125.
[24]王辉.基于自适应模糊神经网络的交通流状态预测[J].交通与计算机,2007(4)：46-49
[25]张敬磊.基于数据挖掘的交通流状态辨识算法研究[D].山东理工大学,2006.
[26]魏善冠.基于支持向量机的高速公路交通流预测研究[D].长安大学,2010
[27]Osuna E, Freund R, Girosi F. An Improved Training Algorithm for Support Vector Machines[A]. Principe J, Gile L, Morgan N. Proceedings of the 1997 IEEE Workshop on Neural Networks for Signal Processing[C]. New York:IEEE,1997,276-285.
[28]Joachims T. Making Large-Scale SVM Learning Practical[A]. Scholkopf B, Burges C, Smola A. Advances in Kernel Method-Support Vector Learning[C]. Cambridge, MA:MIT Press, 1999:169-184.
[29 Platt J. Fast training of support vector machines using sequential minimal optimization[A]. In: Scholkph B, Burges C, Smola A, editors. Advances in Kernel Method-Support Vector Learning[C]. Cambridge, MA:MIT Press,1998:185-208
[30]Keerthi S, Gilbert E. Convergence of a Generalized SMO Algorithm for SVM classifierdesign[J]. Machine learning,2002,46(1):351-360
[31]Nathan H Gartner and Carroll Messer and Ajay K Rathi. Monograph on traffic flow theory [M]. US,1996
[32]May Adolf D. Traffic flow fundamentals [M]. US,1990
[33]姜桂艳,丁同强.交通工程学[J].北京：国防工业出版社,2007.
[34]陈宇峰,向郑涛,陈利等.智能交通系统中的交通信息采集技术研究进展[J].湖北汽车工业学院学报,2010(2)
[35]郑建湖,王明华.动态交通信息采集技术比较分析[J].交通标准化,2009(4)
[36]徐永实,秦福生,邵裕森.浅谈高速公路交通流监控中存在的问题和对策[J].江苏交通,2003(11,12)
[37]李清泉,高德荃,杨必胜.基于模糊支持向量机的城市道路交通状态分类[J].吉林大学学报(工学版),2009.9(Vol.39 Sup.2)
[38]白竹.城市主干路交通异常状态自动判别方法研究[D].长春：吉林大,2006.
[39]Sun X T, Mufioz L, Horowitz R. Highway traffic state estimation using improved mixture kalman fillers for effective ramp metering control [C]. Proceedings of the 42nd IEEE Conference on Decision and Control, Hawaii, USA.2003:3413-3417.
[40]Porikli F, Li XK. Traffic congestion estimation using HMM models without vehicle tracking [C]. IEEE Intelligent Vehicles Symposium, Parma, Italy.2004:188-193
[41]江龙辉.城市道路交通状态判别及拥挤扩散范围估计方法研究[D].长春：吉林大学2007.
[42]刘静,关伟.交通流预测方法综述[[J].公路交通科学,2004,121(13)82-85
[43]王正武,黄中祥.短时交通流预测模型的分析与评价[[J].系统工程,2003,21(6)：97—100
[44]贺国光,李宇,马寿峰.基于数学模型的短时交通流预测方法探讨[[J].系统工程理论与实践,2000,12：51-56
[45]ITS America[EB/OL]. http://www.itsa.org
[46]Brian L. Smith, Billy M. Williams, R. Keith Oswald. Comparison of parametric and nonparametric models [J]. Transportation Research Part C,2002,10 (4):303-321
[47]宫晓燕,汤淑明.基于非参数回归的短时交通流量预测与事件检测综合算法[[J].中国公路学报,2003(1)：82-86
[48]杨兆升,朱中.基于卡尔曼滤波理论的交通流量实时预测模型[J].中国公路学报,1999,12(3)：63-67
[49]张业平,张起森.尖点突变理论在交通流预测中的应用[[J].系统工程学报,2000,15(3)：272-276
[50]贺国光,马寿峰,李宇.基于小波分解与重构的交通流短时预测法[[J].系统工程理论与实践,2002,9：101-106
[51]王晓原,吴磊,张开旺,张敬磊.非参数小波算法的交通流预测方法[[J].系统工程.2005,23(10)：44-47
[52]L. Florio. Neural-Network Models for Classification and Forecasting of Freeway Traffic Flow Stability [J]. Control Eng. Practice,1996,4(2):153—160
[53]朱中,杨兆升.实时交通流量人工神经网络预测模型[J].中国公路学报,1998,11(4)：89-92
[54]杨兆升,谷远利.实时动态交通流预测模型[J].公路交通科技,1998,15(3)：4-7
[55]谭国真,丁浩.广义神经网络的研究及其在交通流预测中的应用[[J].与决策,2002,17(增刊)：777-784
[56]杨兆升,姜桂艳.基于高阶神经网络的城市交通诱导理论模型[J].公路交通控制科技,1998,115(12)：102-106
[57]初连禹,杨兆升.基于神经网络的实时交通流量模糊预测[[J].中国公路学报,998,11(增刊)：78-82
[58]杨世坚,贺国光.基于模糊C均值聚类和神经网络的短时交通流预测方法[J].系统工程,2004,22(8)：83-86
[59]Chang, S.C., Kim, R.S., Kim, S.J.. Traffic-flow forecasting using a 3-stage model[C]. IEEE Intelligent Vehicles Symposium, Proceedings,2000, p 451-456
[60]Yang Licai, Jia Lei, Wang Hong. Wavelet network with genetic algorithm and its applications for traffic flow forecasting[C]. Fifth World Congress on Intelligent Control and Automation,2004, p 5330—5333
[61]Jiang Xiaomo, Adeli Hojjat Wavelet packet-autocorrelation function method for traffic flow pattern analysis[J]. Computer-Aided Civil and Infrastructure Engineering,2004,19(5):324-337
[62]徐岩,石丽坤,王明蕙.基于支持向量机的大学生毕业选择预测研究[J].石家庄学院学报,200810(6)：34-38.
[63]宁朝.基于SVM技术的英文字符识别方法[J].大众科技,2005(8)：160-161.
[64]姜教闻.应用支持向量机的交通流状态预测方法研究[J].西南民族大学学报(自然科学版),2009(4)
[65]王晓原,刘海红,王风群,王晓辉.交通流短时预测理论研究进展[J].交通标准化,2006.
[66]曹劲劲.高速公路与城市快速路结合部交通状态判别及预测研究[D].北京交通大学,2009.
[67]吴慰.短时交通流预测的PSO-PLS组合预测模型研究[D].重庆大学,2009.
[68]殷礼胜.交通流量时间序列混沌特性分析及预测研究[D].重庆大学,2007.
[69]邓毅萍.高速公路路段运行状况评价与分析研究[D].东南大学,2005.
[70]喻丹.基于小波理论的短时交通流预测方法研究[D].长沙理工大学,2008.
[71]刘海红.交通流状态非参数辨识关键理论及方法研究[D].山东理工大学,2006.
[72]范鲁明.基于非参数回归的短时交通流量预测[D].天津大学,2007.
[73]Ren J T, Ou X L, Zhang Y. Research on network level traffic pattern recognition [C]. IEEE 5th International Conference on Intelligent Transportation Systems, Singapore,2002:500-504.
[74]Wu C H, Wei C C, Chang M. Travel time prediction with support vector regression [J].IEEE Transaction on Intelligent Transportation Systems.2004,5(12):276-281.
[75]Scholkopf B, Alex J. New Support Vector Algorithms [J].Neural Computation,2000, 12(5):1207-1245.
[76]Yun S Y, Namkoong S. A performance evaluation of neural network models in traffic volume forecasting[J].Mathematical and Computer Modeling,1998,27(6):293-310.
[77]丁爱玲.基于统计学习理论的交通流量时间序列预测[J].交通与计算机,2002,20(2)：27-30.
[78]Ding A, Zhao X, Jiao L.Traffic flow time series prediction based on statistics learning theory[C].In Proeeedings IEEE 5th Int.Conference on Intelligent Transportation Systems,2002:727 —730.
[79]徐启华,杨瑞.支持向量机在交通流量实时预测中的应用[J].公路交通科技,2005,22(12)：131-134.
[80]蒋刚,肖建.基于核机器方法的城市交通流量实时预测[J].计算机工程,2006,32(17)：48-51.
[81]杨兆升,王媛,管青.基于支持向量机方法的短时交通流量预测方法[J].吉林大学学报(工学版),2006,36(6)：881-884.
[82]姚智胜,邵春福,高永亮.基于支持向量回归机的交通状态短时预测方法研究[J].北京交通大学学报,2006,30(3)：19-22.
[83]姚智胜,邵春福,熊志华.基于主成分分析和支持向量机的道路网短时交通流量预测[J].吉林大学学报(工学版),2008,38(1)：48-52.
[84]张朝元,陈丽.改进的LS—SVM算法及在交通流量预测上的应用[J].昆明理工大学学报(理工版),2008,33(6)：72—75.
[85]刘红军.城市交通监控系统分析与研究[D].武汉理工大学,2003.
[86]姜雪花.城市动态交通流信息采集点布设及交通流短期预测方法研究[D].东南大学,2006.
[87]关伟,何蜀燕.基于统计特性的城市快速路交通流状态划分[J].交通运输系统工程与信息,2007,7(5).