Study On Urban Public Transport Passenger Flow Forecasting Method Considering Geographic Factors

With the development of urban modernization,urban traffic pressure is becoming more and more heavy."Priority development of public transport" has become a consensus,and is also the fundamental way to solve urban traffic problems.In the problem of how to develop urban public transport,it is very important to scientifically and reasonably formulate urban public transport planning plan and optimize management in operation.Bus passenger flow forecasting is a basic work in urban public transport planning and design.It is the main basis for the design of bus routes and stations.It is also an important reference factor for the scheduling of urban public transport operations.However,the existing bus passenger flow forecasting methods are mainly based on historical passenger flow data,and generally do not consider the spatial characteristics of urban road network structure,population density distribution and so on.Although the calculation is simple,it is difficult to improve the prediction accuracy due to ignoring the impact of geographical factors on passenger flow,and it has become increasingly difficult to meet the future development needs of more accurate,efficient,humane and intelligent urban public transport.Therefore,it will become an important research topic to add the relationship between bus passenger flow and urban spatial structure to the bus passenger flow prediction in order to further improve the accuracy and efficiency of bus passenger flow prediction.This paper analyses the research background and current situation of bus passenger flow forecasting.In view of ignoring geographical factors in current bus passenger flow forecasting,an improved segment model for bus passenger flow forecasting is proposed based on urban road structure and population density distribution,so as to improve the correlation between road proximity centrality and bus passenger flow.On the basis of dynamic passenger flow statistics,static urban spatial structure information is added,and urban public transport passenger flow forecasting model considering geographical factors is constructed.BP neural network passenger flow forecasting model is optimized by using double-layer genetic algorithm in order to improve the accuracy and efficiency of passenger flow forecasting.The main work and innovations are as follows:(1)An improved segment model for bus passenger flow prediction is proposed.Aiming at the problems of neglecting spatial scale and bus station distribution in the current study of bus passenger flow and failing to reveal the characteristics of bus passenger flow distribution from the perspective of passenger flow occurrence,this paper studies the relationship between road proximity centrality and bus passenger flow under different road network models based on the calculation method of road proximity centrality.In order to solve the problem of low correlation between the two models,the method of arc point interruption and weighted number of stations is used to improve the line segment model.Compared with the existing methods,the model can effectively reflect the accessibility of urban roads,greatly enhance the correlation between road proximity centrality and bus passenger flow,and provide support for the subsequent bus passenger flow prediction.(2)This paper puts forward a method of urban bus passenger flow forecasting considering geographical factors.In view of the neglect of urban geographical factors in current bus passenger flow forecasting research,static urban geographic data and dynamic historical bus passenger flow data are used as the basis of passenger flow forecasting.BP neural network model is used to study the impact of different geographical factors on the accuracy of urban bus passenger flow forecasting.The experimental results show that adding the static urban geographic data and dynamic historical bus passenger flow data into the model can improve the accuracy of urban bus passenger flow Urban geographical factors can effectively improve the accuracy and efficiency of bus passenger flow forecasting.(3)BP neural network model optimization method based on double-layer genetic algorithm is proposed.Aiming at the problems of imprecise structure design of BP neural network,slow convergence speed of model and low accuracy of prediction results,a double-layer genetic algorithm is designed to optimize the structure and parameters of BP neural network simultaneously.Define the appropriate function that can satisfy both the structure and accuracy requirements,and ensure the unity of genetic algorithm and BP neural network.The optimal solution retention strategy is adopted to reduce unnecessary genetic iteration,and the adaptive probability calculation method is used to improve the efficiency of genetic algorithm and enhance the credibility of the optimal solution.The optimized BP neural network passenger flow prediction model has stable network structure,good learning speed and accurate prediction accuracy.(4)System experiments.This paper designs and implements the urban bus passenger flow analysis and prediction system,integrates the proposed algorithm model and optimization scheme,and verifies the rationality,effectiveness and practicability of the theory and method proposed in this paper.