贝叶斯框架下快速公交的站间运行状态推断

站间行程车速是度量快速公交运营服务和乘客体验的重要指标.由于红绿灯、行人过街等影响,快速公交车速分布往往呈现多峰现象,很难利用单一分布描述.本文建立了贝叶斯框架下基于多相站间车速的高斯混合模型,利用可逆回跳马尔科夫链蒙特卡洛方法推断快速公交站间运行状态,包括状态个数和各状态权重、速度均值和方差.以广州市快速公交为例,展示了该模型可很好地拟合多相车速数据,合理划分快速公交运行状态.发现广州市快速公交在早晚高峰容易衍生出新的运行状态,对于车速低于25 km/h的走廊,其运行状态个数较多,且更容易随时段发生变化,符合实际调查结果.可见本文提出的模型能有效地分析快速公交的速度分布特征和演变规律,验证了该模型的有效性.

Inferring the Travel States of Bus Rapid Transit between Neighboring Stations within a Bayesian Framework

Travel speed is a fundamental measure of operation service and passenger experience for bus rapid transit (BRT). The distributions of travel speed of BRT show multimodal shapes, because of the influence of the traffic lights, the pedestrian, etc. Traditional mathematical models with simple distribution lack the ability to explain these factors that lead to different speed distribution curves. A Gaussian mixture model based on the multimodal travel speed data is constructed under a Bayesian framework in this paper. A reversible-jump Markov chain Monte Carlo approach (RJMCMC) is used to infer the detailed travel states information of BRT, including the number of travel states and the weight, speed and variation for each travel state. A case study was presented in Guangzhou, China. The results show the distributions of the multimodal travel speed are well fit with the model and the travel states are divided properly. New travel state is easily derived during morning or evening peak, and the corridor with the speed lower than 25 km/h had higher number of travel states and the states are easier to change in time of day. The study shows that the model could be used to analyze the distributed features and evolution rules of travel states and proves the availability of the proposed model