最小二乘支持向量机交通事件检测算法  

Traffic Incident Detection Algorithm Based on Least Squares Support Vector Machines

作　　者：梁新荣[1] 裴瑞平[2] 刘智勇[2] 徐建闽[1] 

机构地区：[1]华南理工大学交通学院 [2]五邑大学信息学院,广东省江门市529020

出　　处：《微计算机信息》2007年第02X期196-197,209,共3页Control & Automation

基　　金：广东省自然科学基金资助项目(06300326)

摘　　要：根据高速公路有事件发生时交通流将产生突变这一原理,采用最小二乘支持向量机研究高速公路事件检测问题。阐述了最小二乘支持向量机分类算法,设计了事件检测系统结构,分析了交通事件对交通流的影响规律,并合理地选取了最小二乘支持向量机的输入量。仿真实验表明,最小二乘支持向量机事件检测算法具有检测准确率高、学习速度快等优点,具有良好的应用前景。According to the principle that an incident on a freeway produces a great impact on the traffic flow, the problem of freeway incident detection is studied by using least squares support vector machines （LS-SVM）. The algorithm of the LS-SVM classification is formulated. Then the structure of an incident detection system is designed. The influence of an incident on the traffic flow is ana- lyzed, and the LS-SVM input variables are selected reasonably. Simulation experiments show that the LS-SVM incident detection al- gorithm has such advantages as high detection rate and fast learning ability. It is found to be potentially applicable in practice.

关 键 词：交通工程 事件检测 最小二乘支持向量机 分类 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]