混合自行车交通流速度分布模型

Speed distribution model for heterogeneous bicycle traffic flow

针对电动自行车和普通自行车在非机动车道上混合运行的问题,基于实测数据分析混合自行车交通流速度的基本统计特性.通过对多种影响因素的分析,构建基于高斯混合模型(GMM)的速度分布函数,采用期望最大化(EM)算法对模型参数进行最大似然估计.通过Kolmogorov-Smirnov(K-S)拟合优度检验优化,得到高斯混合模型的最佳组成数.分析不同限速阈值对自行车超速特性的影响.结果表明,利用高斯混合模型能够有效地拟合混合自行车速度.利用三元高斯混合模型能够拟合自由流状态下的速度数据;针对多种交通状态下的数据,须采用五元或六元高斯混合模型进行拟合.

The basic statistical properties of speeds for heterogeneous bicycle traffic flow were analyzed based on the field survey data considering the situation that electric bicycles and regular bicycles ride on the bicycle lane together.A Gaussian mixture model(GMM)for bicycle speed distribution was constructed,and the expectation maximization(EM)algorithm was used for the maximum likelihood estimation of model's parameters through the analysis of various impact factors.The optimal number of components for GMM was determined by using Kolmogorov-Smirnov(K-S)goodness of fit test.Then the effect of different speed limits on bicycles' over-speed percentages was analyzed.Results show that the GMM can fit the field heterogeneous bicycle speed samples well.Three-component model can be used for fitting speed samples under free flow conditions,but five-or six-component model(GMM)should be used under both congested and uncongested conditions.