交通流量变模式辨识的非参数概率变点模型

交通流状态辨识是智能运输系统（ITS），尤其是其先进的交通管理系统（ATMS）和先进的出行者信息系统（ATIS）研究的一个重要内容。以往算法主要集中在交通流预辨识（即交通流预测）和交通流实时辨识（即事件检测或交通流突变检测）的研究上．而忽略对交通流关键参数量变模式的辨认。依据交通流理论，结合非参数变点统计方法，以在淄博市世纪路采获的自由车流数据的研究为例，首次建立对交通模式量变识别的非线性概率变点模型．对变点存在与否进行假设检验。并对变点发生（时间）位置的搜索算法作了详细研究。最后结合实地数据对本方法进行了标定和检验。验证了本方法的有效性和可行性。

A Nonparametric Probability Change-point Model for Traffic Flow Recognition

Traffic flow condition recognition is one of the important issues for Intelligent Transportation Systems research,especially for Advanced Traffic Management Systems and Advanced Traveler Information Systems research.The former (algorithms) were mainly concentrated on traffic flow recognition in advance(namely traffic flow forecasting) and real-time recognition(namely incident detection or traffic flow breakdown detection),but recognition of quantitative change of key parameters of traffic flow was neglected.Based on traffic flow theory and nonparametric statistical method of change-point,a nonlinear probability change-points model to identify quantitative change was firstly established with free flow of Shiji Road of Zibo city as an example,the hypothesis-testing problem of change-points and the algorithm to search the change-points were discussed.Finally,the method was calibrated and tested with the field data to verify the validity and the feasibility of the(theory.)