The passenger flow status identification based on image and WiFi detection for urban rail transit stations

摘要

During the peak hours, the concentration of passenger flow is relatively high for some busy subway lines, if the measures can't be taken in time, more serious accidents may happen, which will influence the social image of the subway. At present, the passenger flow of the key stations is judged mainly by the experience of the staffs, and then the corresponding measures are taken, the errors may be large, and the relevant technical research is urgently needed. First, a data collection device called "the elf of passenger flow-collecting", which integrates high definition camera image acquisition equipment and WIFI probe technology was set up. It can be used to collect the original passenger flow data of congestion points of subway stations. Second, a convolution neural network passenger flow identification algorithm based on deep learning is designed, which is used to estimate the P-0 of stations. Third, because of the error in the video image recognition algorithm, the WIFI probe data acquisition scheme is designed, and the SQL preprocessing assembly for WIFI data processing is established. The noise of WIFI probe is preprocessed, and the flow rate of P-5 based on WIFI probe is obtained. The difference between P-0 and P-5 is defined, and the degree of the difference between P-0 and P-5 is calculated, so the final passenger flow P(6 )can be obtained. Finally, the Songjiang University Hall Station of Shanghai Metro line 9 was taken as an experimental analysis object, the high definition camera and WIFI probe are set up on the spot, the passenger flow video data and the WIFI data are collected synchronously, so the real-time passenger flow in the station's internal position is estimated, and the accuracy is corrected, meanwhile the passenger flow early warning of the station position is obtained. An emergency response plan based on passenger flow early warning level is proposed, and the flow chart of passenger flow density inside Songjiang University hall station is drawn. The construction of the equipment platform and the identification and correction methods of passenger flow are of good practical guiding significance for the Metro to run safely. (C) 2018 Elsevier Inc. All rights reserved.