基于交叉验证支持向量机算法的交通状态判别研究

PDF全文下载：2017010105

高林，盛子豪，刘英

(青岛科技大学 自动化与电子工程学院，山东 青岛266042)

 摘要：传统交通状态判别算法存在准确率低和速度较慢的缺点。将支持向量机(SVM)算法应用于交通状态监测中，并且以径向基函数（RBF）作为核函数，在SVM训练过程中采取交叉验证的方法，即将训练样本分成多个子集，再通过每次保留其中一个子集作为检测样本进行反复训练，获得RBF所需的σ值和惩罚系数C。通过对某主干路上某时段的数据验证和检测表明，基于交叉验证SVM算法具有较高的准确率和速度，可以有效地将其运用到实际中。

 关键词： 支持向量机；交叉验证；交通状态判别

中图分类号：U 491文献标志码：A

引用格式：高林，盛子豪，刘英.基于交叉验证支持向量机算法的交通状态判别研究［J］.青岛科技大学学报(自然科学版), 2017, 38(1): 105108.

GAO Lin, SHENG Zihao, LIU Ying.Identification of traffic state based on cross-validation SVM［J］.Journal of Qingdao University of Science and Technology（Natural Science Edition), 2017, 38(1): 105108.

 Identification of Traffic State Based on Cross-Validation SVM

 GAO Lin, SHENG Zihao, LIU Ying

 (College of Automation and Electronic Engineering,Qingdao University of Science and Technology, Qingdao 266042, China)

Abstract: Because traditional traffic identification algorithm has the shortcomings of low accuracy and slow speed, we applied SVM (support vector machine) which its learning procedure is based on the method of crossvalidation into detecting traffic state with the RBF (radial basis function) core. The cross-validation SVM separates the training samples into several subsets, and each subset was tested by the classifier created by training other subsets. Then the best parameters of SVM and RBF,σand C were gained. In the end, we used the data from a period of a main road as training samples to test SVM based on cross-validation, and the experimental result shows that, with this method, we can detect traffic state fast and precisely. In addition, this method can be applied into practice effectively.

 Key words: support vector machine; cross-validation; identification of traffic state

 收稿日期：   20160202

基金项目：山东省自然科学基金项目（ZR2014FL018）;青岛科技大学博士启动基金项目（010022530）.

作者简介：高林（1976—），男，副教授.

文章编号：16726987（2017）01010504；DOI：10.16351/j.16726987.2017.01.018