Research On Pedestrian Incident Automatic Detection Technology Of Urban Rail Transit Hub

Channel whose function is to make passengers transfer and get in/out transit hub is one of the infreatructures in urban rail transit hub. Channel is always the bootleneck area of transit hub because its characteristic of time distribution is local concentration and global expanding. Therefore operation level of urban rail transit hub is decided by though capacity of channel. In modern society, abnormal behavior is always influenced on surrounding pedestrians caused by social activities, and it will influence on traffic function of urban rail transit hub. Once incident occurs in channel which is the key note transit hub, it will lead to channel capacity dropping and congestion. Service level of channel will be influenced serviously. In some peak periods, operation level of transit hub will even influenced by congestion of channel because of chain effect. Therefore, to monitor passenger flow in real time, to find incident timely and to organize pssernger flow reasonably are the important guarantee to make transit hub operate efficiently and safely. For these above reasons, this paper study on pedestrian incident automatic detetion technology of urban rail transit hub to make up defectiveness of automatic detection technology of traffic state.This paper is based on The Study on Key Technology of Transfer Information Sensing and Co-Scheduling, Science and Technology Project of Ministry of Transport. Purpose of this paper is to ensure that channel operate efficently and safely. Content of this paper includes: analysis of characteristic of channel passenger flow, channel pedestrian detection method based on machine vision, channel incident simulation model, analysis that incident impacts on channel passenger flow and channel incident automatic detection algorithm.Analysis of characteristics of channel passenger flow is the base of channel channel incident simulation model. Channel is claffied in the paper based on the analysis of temporal distribution characteristic of channel. Then this paper studies on how channel type make influence on channel service level.Channel pedestrian detection method based on machine vision can provide long-term raw data, so it can ensure the study of channel incident automatic detection algorithm conveniently. This paper proposes a automatic detection method of channel passenger flow which is suit for Shor-shot and nondirect downward view. This method use removing false detection rectangel algorithm to improve detecting procedure, use templae merging algorithm to improve tracking procedure and use threshold curve to improve people counting procedure.Channel incident simulation model is the base of analysis of channel passenger flow characteritic under incident circumstance and channel incident automatic detection algorithm. This paper proposes a simulation model considering pedestrian gathering. Pedestrian is classified to three types including “unaffected”, “stopping” and “onlooing” to simulate pedestrian behavior when incident occurs. This paper is also take model validation and sensitivity analysis.Study on how incident impacts on channel passenger flow is the base of studying the relationship between incident type and incident automatic detection algorithm. This paper develops channel incident simulation and analysis system. Analysis that how different types incidents impact on channel passenger flow through designing indexs of mutation rate and average change rate of traffic parameter using this system.Channel incident automatic detection algorithm is the core of channel incident automatic detection technology. This paper designs channel incident automatic detection algorithm under ideal circumstance and considering detection error based on channel incident simulation model, study on how incident imapcts on channel passenger flow and study on characteristic of basic data and key parameter. Finally, validity of the algorithm has been verified.In this paper, some contents including analysis of characteristic of channel passenger flow, channel pedestrian detection method based on machinge vision, channel incident simulation model, analysis that incident impacits on channel passenger flow and channel incident automatic detection algorithm are studied on the basis of summarizing research status both at home and abroad. Study result of this paper will strengthen the technology of monitoring and detecting pedestrian in urban rail transit hub, improve the intelligent level of administration and reduce the security risk caused by incident.