基于支持向量机的高速公路事件检测

针对交通领域中的事件检测(无事件模式和事件模式)模式识别问题,描述了支持向量机(SVM)支持的基本方法,建立了基于线性(linear function)、多项式(polynomial function)和径向基(radial basis function)3种核函数的事件检测SVM模型。采用高速公路路段I-880线圈数据集和事件数据集验证模型,结果发现:无论对于向北、向南或混合方向的事件检测,在3个SVM模型中,SVM(P)检测效果最好,SVM(L)最差。SVM算法具有避免局部最小,实现全局最优化,更好的泛化效果的优点,是高速公路事件检测的一种很有潜力的算法。

Incident Detection on Highway Based on Support Vector Machines

Directing at the problem of mode identification of incident detection in transportation field, this paper describes the basic methods of support vector machine (SVM) in incident detection, and establishes three SVM models of incident detection respectively based on linear function, polynomial function and radial basis function. An illustration was conducted to validate these models by taking the data from the I-880 section on highway, the result shows that SVM(P) is the best and SVM(L) is the worst among the three SVM models. SVM is an effective algorithm in incident detection and superior to neural network in avoiding local minimum and achieving global optimization.