基于小波分析和非参数回归的交通流组合预测方法  

Traffic Flow Prediction Based on Wavelet Analysis and Nonparametric Regression

作　　者：窦慧丽[1,2] 吴志周[2] 

机构地区：[1]浙江交通职业技术学院运输管理学院,浙江杭州311112 [2]同济大学道路与交通工程教育部重点实验室,上海201804

出　　处：《重庆交通大学学报（自然科学版）》2011年第4期807-811,共5页Journal of Chongqing Jiaotong University(Natural Science)

基　　金：国家自然科学基金重点项目（70631002）;国家“863”计划项目（2008AA11Z205）

摘　　要：针对实际交通系统时变复杂的特征和交通流变化的不确定性，基于模式识别的思想，提出了一种小波分析和K近邻非参数回归相结合的交通流组合预测模型。模型首先应用小波分析理论，对原始交通数据进行了消噪处理，使消噪后的数据更能反映交通流的本质及变化规律；然后采用K近邻非参数回归模型对交通流进行预测；最后根据该模型，结合实测交通流数据进行了预测实验分析。结果表明：该方法具有较高的预测精度，可用于交通流的实时动态预测。According to the complexity of the field traffic system and the uncertainty of the traffic flow, a traffic flow predic- tion model which combined an algorithm of wavelet analysis with nonparametric regression was established based on the thoughts of pattern recognition. First, theory of wavelet analysis was applied to eliminate the effect of noise on the real-time traffic data, which can reflect the essence of traffic flow. Then with the worked traffic data, K-nearest neighbor nonparamet- ric regression was used to predict traffic flows of the next time period. Finally, according to the proposed algorithm, the prediction analysis was carried out. The results indicate that the model has a finer precision and can be used for the real- time dynamic traffic flow prediction.

关 键 词：交通流预测 小波分析 消噪 K近邻非参数回归 

分 类 号：U491[交通运输工程—交通运输规划与管理]