Safety Status Identification Of Passenger Flow In Metro Station Based On Video Image

With the rapid development of the urban rail transit,the large passenger flow has gradually become a normal phenomenon in the daily operation of the metro.As an important gathering and distributing place of the passenger flow organization,the metro station is often built underground.Once an accident occurs,it is difficult to achieve safe and efficient evacuation of passenger flow.Therefore,based on the video image data of the metro station,this paper studies the passenger flow safety status identification.The research plays an important role in accurately grasping the real-time information and changing status of passenger flow in metro station,efficiently organizing and managing passenger flow in station,and improving the operation efficiency.Based on the passenger flow data of Guangzhou Metro Line 2,this paper analyses the spatial and temporal characteristics of the passenger flow in metro stations from macro and micro aspects.Taking the safe operation of metro stations as the starting point,this paper puts forward the definition of the passenger flow safety status in metro stations,and qualitatively analyses the factors affecting the passenger flow safety in metro stations.The parameters needed for identification of passenger flow safety status and the information collection area of the station are selected.In order to realize the real-time detection of passenger flow in metro stations,this paper proposes a method of identification of passenger flow in metro stations based on YOLO v3.This method takes advantage of the outstanding effect of YOLO V3 in the target detection,transforms USC pedestrian data base into YOLO v3 training label format,and obtains the YOLO network model suitable for metro stations by training the pedestrian image of USC pedestrian data base.Taking Nancun Wanbo Station of Guangzhou Metro Line 7 as an example,the monitoring video is used as input vector of the YOLO v3 model to identify the passenger flow at the platform and staircase.The test results show that the average false detection rate of passenger flow detection method based on YOLO v3 is 19.19%,and the average detection speed is 26 fps,which can realize the real-time detection of the passenger flow.In order to obtain the density and velocity parameters of passenger flow in metro stations,the camera calibration parameters are obtained based on the pinhole imaging principle and perspective transformation method.The measurement method of monitoring plane area is put forward.According to the pedestrian coordinates obtained from the passenger flow identification in Chapter III,the density and velocity parameters of passenger flow in metro station are identified through the camera projection model.Based on the criteria of service level of Metro station,the level of passenger flow safety status is divided into level 1(very safe),level 2(safe),level 3(general),levels 4(dangerous),level 5(very dangerous)and level 6(very dangerous).Based on the existing research results,the evaluation parameters are selected and the threshold values of parameters for each level are set.The assessment model of safety state based on PNN is constructed and related parameters are calibrated.A simulation station of safety state of metro passenger flow is built based on the Anylogic software.Six different scenarios are simulated by setting relevant parameters of station facilities and passenger flow,and the passenger flow parameters under six scenarios are obtained.And taking the parameters as an input of PNN network model,the passenger flow safety state of metro stations is simulated.