Analysis And Prediction Of Passenger Flow Characteristics Of Urban Rail Transit Based On ARMA Model

Passenger flow is an important part of urban rail transit system.The analysis and prediction of passenger flow characteristics are often the basis of urban rail transit planning,construction and operation.With the acceleration of urbanization in China,some urban rail transit has formed a network structure.As an important part of modern urban public transport,urban rail is favored by more and more cities and citizens for its punctuality,high efficiency,large traffic volume,clean environmental protection,energy saving and other characteristics.Based on the ARMA model in time series analysis,this thesis analyzes and explores the passenger flow characteristics of urban rail transit,establishes a suitable passenger flow prediction model,and verifies the scientificity and feasibility of this passenger flow prediction model with the historical passenger flow data of Shanghai Metro Line 4.In order to facilitate the establishment of subsequent passenger flow prediction model,this thesis first analyzes the passenger flow characteristics of urban rail transit.Based on the historical data,this thesis analyzes the characteristics of urban rail transit passenger flow,including the characteristics of year,month,day,holiday and normal day,obtains the periodicity,growth,similarity and other characteristics of urban rail transit passenger flow,analyzes the causes of the characteristics combined with the actual situation,and understands the change rule of urban rail transit passenger flow.Then,according to the similarity characteristics of daily passenger flow,cluster analysis is carried out.It is found that the passenger flow data of urban rail transit has great similarity in the same day of the week,including the similarity of passenger flow and the similarity of passenger flow change law,especially the similarity and internal difference of weekend passenger flow and weekday passenger flow.The clustering analysis of passenger flow can further analyze the daily passenger flow characteristics of urban rail transit,and improve the correlation between data,so as to better fit the ARMA prediction model.According to the clustering results,the passenger flow data was predicted by clustering,and the prediction results were obtained.Compared with the prediction results without clustering,the results show that the prediction accuracy of passenger flow after clustering is higher than that of passenger flow prediction without clustering.Finally,considering that the nonlinear part of daily passenger flow data affects the accuracy of ARMA model,ARIMA model and RBF model are combined to predict daily passenger flow.Firstly,ARIMA model is used to model and predict historical data,and then a prediction result and prediction error are obtained;secondly,RBF neural network model is used to fit the ARIMA prediction error,and the linear and nonlinear characteristics of passenger flow are fully considered;finally,the final prediction result is obtained by combining the two prediction results.At the same time,a case study was conducted to verify the prediction results of single ARIMA model and RBF model.The results show that the combined model proposed in this paper can fit the passenger flow data more comprehensively,and the prediction accuracy is better than that of single model.