短时交通流预测方法研究  

Research on short time traffic flow forecasting method

作　　者：郭牧[1] 孙占全[1] 潘景山[1] 徐梅[1] 

机构地区：[1]山东省计算中心,济南250014

出　　处：《计算机应用研究》2008年第9期2676-2678,共3页Application Research of Computers

基　　金：山东省电子发展专项基金资助项目(2006R00042)

摘　　要：交通流预测是智能交通系统的热门研究课题,如何构建合适的预测模型并选择合适的预测变量是交通流预测的关键。利用相关分析法来确定交通流预测变量,将选择的预测变量输入到非线性回归支持向量机,通过样本训练进行交通流预测,最后通过交通实例分析来验证该方法的有效性。Traffic flow forecasting is a popular research topic of intelligent transportation systems. How to build suitable forecasting model and select suitable forecasting variable is a key problem. This paper took use of correlation analysis to select the close relate forecasting variables and input the variables to nonlinear regression SVM. The SVM was used to forecast traffic flow through training with history traffic flow data. Illustrated the method＇ s efficiency through analyzing the practical traffic data.

关 键 词：相关分析 支持向量机 交通流预测 智能交通 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]