Crowded treading warning system for urban rail transit stations based on video detection technology

Abstract

Abstract Based on the rapid development of rail transit, it have a greater hidden danger of congestion and trampling, this paper proposes the construction of an urban rail transit station congestion and trampling warning system based on video detection technology, which uses video detection technology to collect the passenger flow, velocity and density data of the bottleneck channel in rail transit stations, then build and train neural network model to predict the next time passenger flow density combine with the three-parameter relationship of the traffic flow, and use the SPSS software used to fit and analyze the passenger flow parameters. So the limit value of the flow-density polynomial model is predicted by pedestrian flow mutation model and used to realize the definition of the critical passenger flow density model based on the catastrophe theory. On this basis, this article compares the real-time predicted value of passenger flow density to it and get the analysis result, then set the corresponding early warning coefficient, early warning level and early warning measures, so it can truly realize the real-time warning of congestion and trampling in rail transit stations based on passenger flow density prediction.