Research On Vehicle Congestion Identification Method Of Urban Rail Transit

In recent years,with the rapid advancement of urbanization in China and the rapid development of urban rail transit system,the increasing scale of line network brings new challenges to the daily operation and safety management of urban rail transit.Subway vehicle congestion is a kind of psychological feeling that passengers consider comfort and safety in the process of riding,and it is an important index to measure the service quality and service level of urban rail transit system.The real-time detection of subway vehicle congestion can truly reflect the passenger flow state of the line network,provide more comprehensive passenger flow information at the section,and provide a basis for the real-time scheduling and passenger flow control of urban rail transit.This paper studied the problem of vehicle congestion in urban rail transit.Firstly,it discussed the connotation of vehicle congestion in urban rail transit.It mainly sorted out and summarizes the existing evaluation standards of subway vehicle congestion,and then analyzed the influencing factors of subway vehicle congestion on this basis,and focused on the qualitative and quantitative analysis of the relevant characteristics of subway vehicle congestion.Considering that the subway vehicle monitoring video can provide the real-time passenger flow data of the line network,paper proposed a method of vehicle congestion identification based on video analysis.This method,based on the extraction and analysis of surveillance video images from subway trains,using the convolution neural network model to identify the congestion level of trains.It established the passengers data by videos,and proposed a fast congestion identification method based on the performance of on-board computer,which used the convolution neural network with cascade structure.The experiment results showed that the proposed method had a fast detection speed and can meet the real-time requirements in practical applications.The accuracy of three-level congestion degree experiment was 98%,and the accuracy of four-level congestion degree experiment was 87%.Finally,the author analyzed the problems existing in the congestion recognition of twodimensional image.In order to solve the problem of two-dimensional image in congestion recognition,a vehicle congestion recognition method based on three-dimensional lidar was proposed.In this paper,rs-lidar-16 Li DAR was used to collect three-dimensional point cloud data of urban rail transit vehicles.Then based on the analysis and processing of a large number of point cloud data,a feature description method of subway vehicle congestion,which based on point cloud reflection intensity was proposed.At last,the paper used a machine learning method which based on the feature description to realize the recognition of subway vehicle congestion.The experimental results showed that the identification accuracy of 20 human precision units was100%,and that of 10 human precision units was 98.31%.Moreover,with further subdivision of the accuracy,the extracted lidar point cloud feature parameters still had the ability to distinguish.