通道行人集聚型异常事件自动识别算法设计

Design of Automatic Identification Algorithm for Pedestrian Clustering in Channel

为了对城市轨道交通枢纽通道内的集聚型异常事件进行合理的疏导和客流组织,保障城市轨道交通枢纽的安全、高效运行,本文提出了一种通道内行人集聚型异常事件的自动识别算法。该算法首先通过对通道客流基础数据平稳性和突变性的分析,创建了一种兼具平稳性和突变性特征的新数据类型,然后基于双截面客流数据设计了自动识别算法的关键参数—偏移空间差值。最后通过对关键参数变化特征的分析,建立了通道行人集聚型异常事件自动识别算法。仿真试验结果显示:该算法的检测精度为100%,反应时间均值为65 s,表明该算法对通道行人集聚事件有极强的自动检测能力和较短的反应时间。

In order to carry out reasonable guidance and passenger flow organization in the traffic hub channel of urban rail transit,ensure the safe and efficient operation of urban rail transit hub,we put forward an algorithm that can recognize the abnormal events of crowds gathering in the transfer channel automatically.Basic information like stability and mutability of pedestrian volume is analysed firstly,creating a new type data set characterized by stability and mutability based on the calculated result,and then the key parameterdifference of space offset of automatic identification algorithm is designed based on the double-section pedestrian volume,and variation characteristics analysis of the key parameter will help to establish the algorithm for automatic identifying crowds gathering abnormal events. The simulation experiment result shows that the detection accuracy of the algorithm is 100%,and the reaction time is 65 s,which shows that the algorithm has a strong automatic detection ability and a shorter reaction time for the pedestrian clustering events.