基于非参数回归的快速路行程速度短期预测算法

基于北京市快速路上的检测器所采集的历史数据，经过数据筛选，剔除判别，小波滤噪平稳处理，聚类分析等过程，建立了交通状态演变系列的历史样本数据库。基于所构建的历史数据库，通过数值试验，确定了状态向量、距离匹配原则，K近邻值等参量，构建了一种基于K近邻的非参数回归短时交通预测模型，实现了对路段行程速度的短时预测。最后，利用随机选取的历史数据系列对预测模型的精度进行了检验。结果表明，预测算法的精度可以达到90％以上，可以很好地满足ITS应用系统对于交通预测数据的精度要求。

Non-parametric Regression Model Based Short-term Prediction for Expressway Travel Speed

A promising K-nearest neighbor nonparametric regression forecasting model based on typical historical database was developed.The research relied on the historical traffic flow data which was collected by the detectors installed on the expressways of(Beijing.)A series of typical historical databases were established through data filtering,Wavelet analysis and clustering analysis.Based on the databases,the state space,forecasting functions,and neighbor selection criteria are confirmed by means of numeric experiments.Then,the K-NN(nearest neighbor) nonparametric regression short-term travel speed forecasting model was developed for predicting average segment speeds up to 6 minutes into the future.In the end,the accuracy of the prediction model was tested by utilizing several randomly selected historical data series.The results indicate that the forecasting model has a high degree of accuracy above 90%;it is proved that under the support of real-time data,the prediction algorithm can meet with the requirement for Intelligence Transportation System(ITS) applications.