Research On Short-Term Passenger Flow Prediction Method Of Urban Rail Based On Mode Decomposition And Integrated Learning

With the advent of the era of intelligent transportation,artificial intelligence technology is widely used in the transportation industry.In the field of urban rail transit related research,researchers have achieved a large number of research results using artificial intelligence technology.Urban rail transit passenger flow forecasting can provide reliable data support for traffic management departments and operating units,so that effective passenger flow control measures can be taken to optimize urban rail transit operations and improve passenger travel experience.This paper takes the short-term passenger flow prediction of urban rail transit as the research object,extracts and processes AFC passenger flow data,and analyzes the spatial and temporal distribution characteristics of passenger flow in the rail transit system and the related factors affecting passenger flow.On the basis of integrated empirical mode decomposition(EEMD),three improved EEMD methods are proposed to make the passenger flow forecasting model more adaptable to passenger flow data,a short-term passenger flow combination forecasting model for urban rail transit stations is established,and the improved model is verified.The main research content and conclusions of this paper are as follows:(1)On the basis of the AFC passenger flow data of the Beijing Subway,the missing values and outliers are processed to construct an initial passenger flow data set.The time-space distribution characteristics of passenger flow in the rail transit system were excavated from three aspects: passenger flow time distribution characteristics,spatial distribution characteristics,and station passenger flow distribution characteristics.The time distribution of passenger flow in and out of rail transit stations is uneven,and urban rail transit stations are divided into five types.At the same time,it analyzes the influencing factors of short-term passenger flow of urban rail transit,including line network structure,station operation time and passenger flow service,and analyzes its influencing mechanism.(2)On the basis of the EEMD method,three EEMD improvement methods are proposed,and the REEMD-LSTM model,the MEEMD-LSTM model and the HEEMD-LSTM model are respectively constructed in combination with LSTM,from activation function selection,optimization algorithm selection,and overfitting problems To solve the problem of model parameter setting in three aspects,using the actual passenger flow data of Fuxingmen Station of Beijing Subway,experiments were carried out on the three short-term passenger flow prediction models of urban rail transit based on the improved EEMD method,and the experimental results were compared and analyzed to verify Three improved EEMD methods can effectively improve the prediction effect of the passenger flow prediction model,and through comparative experiments,it is found that the prediction accuracy of the HEEMD-LSTM model is the highest.(3)In order to further improve the prediction accuracy and robustness of the model,an urban rail short-term passenger flow prediction model based on improved EEMD and ensemble learning was constructed,and a structural framework of the prediction model based on improved EEMD-Stacking ensemble learning was proposed,combining XGBoost,Light GBM and The LSTM model is used as the base model,and the LR model is used as the meta-model,and the Stacking integration method is used for fusion.Taking the inbound passenger flow of Beijing West Railway Station of the Beijing Metro as an example,the prediction performance of the improved EEMDStacking integrated learning model is analyzed through the overall prediction effect of the model and the comparative analysis of the prediction models,and its use in predicting short-term passenger flow of urban rail transit is verified.and compared and analyzed the operation efficiency and prediction performance of each prediction model for the inbound passenger flow of Beijing West Railway Station and Fuxingmen Station,providing model selection reference for the rail transit operation department in passenger flow prediction.