Passenger Flow Prediction Of Conventional Buses Around New Urban Rail Transit Lines Based On LSTM Model

With the acceleration of the country's urbanization process and the proposition of the "transit priority" strategy,the development of urban public transportation has become increasingly important.Both conventional public transport and urban rail transit are important components of urban public transport.The interaction of the two modes of transportation,especially when the new urban rail line is opened,will have a complex impact on the passenger flow of conventional buses around it.The distance between the conventional bus station and the urban rail station,and the positional relationship between the conventional bus line and the urban rail line will affect the passenger flow of the conventional bus around the urban rail.If it is possible to reasonably predict the changes in the passenger flow of conventional buses in the vicinity of the new urban rail line before it is opened,and dispatch and optimize in time according to the prediction results,the utilization efficiency of urban public transportation resources can be improved.In the context of the opening of the new urban rail line,this paper builds a model based on the LSTM neural network by using conventional bus passenger flow data before and after the opening of the Beijing Metro Line 6 West Extension Line and the South Section of Line 8 through machine learning and statistical analysis methods After the opening of the new rail line,the prediction model of the passenger flow of the surrounding conventional buses.Combined with the reconstruction project of Beijing Metro Line 13,it predicts the changes in the regular bus passenger traffic around the 13-A section of the new urban rail line after its transformation,providing a reference for the bus company's operation and scheduling.The main research contents are as follows:(1)Carding data of conventional bus IC card swiping and analysis,according to the characteristics of the data to propose a targeted cleaning program,using the Pandas module for data cleaning and data fusion,to obtain regular bus line and station passenger flow data.(2)Sort out and summarize the information of conventional bus lines and stations along the new urban rail line,and analyze the passenger flow of the conventional bus lines and stations around the new urban rail line.Mining its passenger flow characteristics,clustering passenger flow distribution characteristics using K-Means algorithm,and analyzing the classification of conventional bus lines and stations.(3)Establish a passenger traffic prediction model for conventional public transit under the new urban rail line based on LSTM neural network.Based on the classified passenger data,train the neural network model and tune the model parameters.Aiming at the problem of low model prediction accuracy,an improved softplus?mv excitation function is proposed to improve model prediction performance.The MAPE function is used to quantitatively evaluate the effect of the model and analyze the evaluation results.(4)Based on the trained LSTM neural network model,with the background of the Beijing Metro Line 13 reconstruction project,the conventional bus lines and stations around the new line of the 13-A section of the reconstruction are predicted after its new urban rail line is opened.The change of passenger flow provides a reference for the bus company's dispatch optimization.