Research On Forecasting Method Of Short Term Passenger Flow Based On Bayesian Theory

In the increasingly competitive passenger transportation market,the railway departments need to better understand the changing trend of passenger flow in order to reasonably allocate resources,flexibly respond to the dynamic market,and improve operational efficiency.Short-term passenger flow forecasting is an important means to master changes in the passenger flow of the railway and is the basis for scientific management of railway transport enterprises.With the growth of time,the railway ticketing system has accumulated a wealth of historical data,and how to use the existing data resources for passenger flow analysis and forecasting has increasingly become the focus of railway ticketing marketing and information technology departments.This paper mainly studies the short-term passenger flow forecasting problem based on Bayesian theory,and aims to use the historical passenger flow data and ticket data during the pre-sale period to predict the passenger flow scale at the time of departure.First,it reviews and summarizes the existing railway passenger flow forecasting methods and explains their application status and limitations.Then it briefly introduces the basic principles of passenger flow forecasting,the choice of methods,as well as the principles,advantages and disadvantages,and application areas of Bayesian theory,and explains the reasons for the use of Bayesian theory for passenger flow forecasting.Besides,it analyzes the characteristics of passenger flow from the departure time,travel duration,holidays and other dimensions to find possible influencing factors related to passenger flow.Then,this paper sets out a passenger flow forecasting model based on Bayesian theory by using historical passenger flow data and pre-sale period ticket sales data separately.At the same time,it explores how to reduce the recalculation of historical data in the face of constantly updated data.At the end of the paper,it verifies the feasibility of the prediction model through real data of the Beijing-Shanghai high-speed rail,and compares the models with the time series methods and neural network method commonly used in the prediction field.