Research On Short-term Passenger Flow Forecasting Method Of Urban Rail Transit Based On SVM

Urban rail transit has several characters as below:safety,fast,large volume,convenience etc,which can effectively alleviate urban traffic congestion and meet people's travel needs,so it has gradually become an important part of the urban transportation system.Therefore,it is quite necessary to analyze historical passenger flow data and construct a suitable passenger flow forecasting model to predict passenger flow.Accurate short-term passenger flow forecasting is of great help to improve urban rail transit operations and management and service quality.Therefore,the following work has been researched about the short-term passenger flow forecast of urban rail transit.(1)Firstly,this paper analyzes rail transit passenger flow,mainly analyzes the influencing factors of passenger flow,the daily and weekly changes of network passenger flow and station passenger flow,so it can obtain the potential passenger flow characters of urban rail transit.(2)The passenger flow forecasting methods commonly used in urban rail transit are researched.This paper analyzes the advantages and disadvantages of various passenger flow forecasting methods,point out the applicability of support vector machine model in passenger flow forecasting of urban rail transit,and finally choose support vector machine model to forecast the short-term passenger flow in urban rail transit.(3)The basic principle,parameters and parameter selection method of support vector machine method are described.The genetic algorithm is used to optimize the support vector machine model,then the improved passenger flow prediction model is constructed using genetic algorithm to optimize penalty parameter,insensitive loss function parameter and kernel parameter in support vector machine model.(4)Based on the historical data of Fenhu Road Station of Suzhou Rail Transit,the input part of the model is determined by clustering method and correlation analysis,and the passenger flow prediction model of penalty parameter,insensitive loss function parameter and kernel parameter in support vector machine model is established step by step based on genetic algorithm.The results of different prediction methods are compared to verify the improvement of the improved passenger flow prediction model.