基于高斯混合模型的道路交通状态特征辨识方法

Method for Identification of Roadway Traffic Characteristics Based on Gaussian Mixture Models

针对交通流流量、占有率和速度3个基本参数,提出一种基于高斯混合模型的道路交通状态特征辨识方法.利用Expectation-Maximization(EM)算法实现对混合模型的参数估计,结合Bayes方法和柯尔莫哥洛夫-斯米尔诺夫检验进行混合模型的选择.应用北京市实际道路交通流参数数据对所提出的方法进行了验证分析,取得一些具有实际应用意义的特征指标,并就此方法应用中的一些问题进行了讨论.

A method is developed in this paper, based on Gaussian mixture models （G models）, to identify roadway traffic characteristics after an analysis of three basic parameters ： traffic flow volume, occupancy and speed. The parameters of G modets are estimated by Expectation-Maximization （EM） algorithm, and Bayes criteria together with Kolmogorov-Smirnov test are used for the choice of fitting G models. The application of the proposed method to the analysis of freeway traffic flow data in Beijing achieves some characteristic indexes of practical significance.