Research On Passenger Flow Estimation Method Of Subway Station Based On Data Fusion

With the economic development,the demand for urban transportation rises sharply.Rail transit has become the first choice for more and more people because of its large volume,small area,safety,convenience,punctuality and efficiency.The passenger flow of rail transit is rising rapidly,which puts forward higher requirements for operation management.Therefore,how to detect and evaluate passenger flow accurately and in real time has become an important problemIn view of the problems of poor accuracy and poor real-time performance of a single passenger flow detection system,this paper cuts in from the perspective of data fusion to obtain the estimated value of passenger flow in the subway station,and establishes a passenger flow estimation model for the subway station.Through this model,the passenger flow estimate value in the subway station can be obtained more accurately,so as to provide a reference basis for passenger flow control for the subway operator.The main research contents of this article include:(1)Analyze multi-source heterogeneous data in subway stations and perform feature value extraction operations.After analysis and selection of AFC data and surveillance video data as the main research data,the concept of swiping card entry(out)station rate is proposed for AFC data.From the video data,passenger flow status such as regional passenger walking speed,regional passenger density and regional passenger number can be extracted parameter.Afterwards,the data is unified in time and space and the data is cleaned to reduce errors.(2)Establish a passenger pit stop model based on data fusion.Through the analysis of the passenger flow lines of the subway station,it explains which bottleneck areas in the subway station,and explains the determination method of the special point area in one step.Based on the analysis of the subway station inbound flow lines,a passenger inbound model based on data fusion was established and solved using BP neural network to obtain the estimated passenger flow in the inbound area.(3)Establish a passenger flow estimation model for subway stations.Based on the analysis of passenger outbound behavior and transfer behavior,the passenger outbound model and the transfer model are established,and the passenger flow estimates in the outbound streamline area and the transfer streamline area can be obtained.Based on the passenger inbound model,outbound model and transfer model,the subway station passenger flow estimation model is proposed to obtain the passenger flow estimated value in the subway station.Finally,a solution to the problem of blind spots and coincidence in video data is proposed.(4)Take the Octagonal Amusement Park subway station as an example,and use the data of this station to verify the above model.Through the analysis of the passenger flow of the station,the special point area in the station is determined.After that,the passenger flow data substituted into the morning peak of a certain working day was used to verify the model.The average error between the true value and the predicted value was 5.7%.The results obtained were good,which proved the scientificity and effectiveness of the method in this paper.