Study On The Forecast Of Urban Rail Transit Passenger Flow Based On Grey Model

With the accelerating process of urbanization in our country,the rapid economic development,the constant expansion of the scale of urban population,ground traffic congestion,environmental pollution and other issues have become increasingly prominent,so vigorously develop in urban rail transit system as the main mode of public transport potential in.Now many large and medium-sized cities in China are in the vigorous development of urban rail traffic system.Passenger flow,is the basis for urban rail transit planning,construction and operation of the various stages.The accuracy of the results of passenger flow prediction will directly affect the project investment and operation efficiency of urban rail transit.However,a large number of facts indicate that the passenger flow prediction is often different from the actual passenger flow.Firstly,considering the city rail traffic flow influence factors of diversity,complexity and randomness of the premise,based on the analysis and comparison of the existing passenger flow forecasting model,finally chose the gray model for the passenger flow forecasting model.According to the different distribution characteristics in time and space of city rail transit passenger flow,so we select the average daily traffic of Tianjin urban rail transit Line 3,among 2016-2020 year in November the ordinary working day(Monday to Thursday),special working day(Friday)and weekend(Saturday and Sunday)in three different conditions as prediction object.This paper gives a detailed description on the basic idea,the process of modeling and the accuracy test of grey prediction model,an improved model based on grey prediction model is put forward,which is the grey metabolic model.With an example,it is proved that the grey prediction model has a very high precision in forecasting the short-term passenger flow of rail transit.Finally,the new model was adopted to predict the average daily traffic of Tianjin urban rail transit Line 3 in 3 different cases in November among 2016-2020 year.