Analysis Of Bus Trip Characteristics And Forecasting Methods Of Bus Passenger Flow

With the acceleration of urbanization,the number of motor vehicles has increased sharply,and traffic congestion has become increasingly serious.Research experience shows that giving priority to the development of public transport and formulating a scientific and reasonable public transport operation scheduling scheme are the key measures to solve the problem of urban traffic congestion.Bus passenger flow information is the basic information for bus operation management departments to plan and make dispatching plans.Only when bus system administrators accurately and efficiently collect bus passenger flow information can they provide guarantee for the formulation of plans.At present,there are many manual survey methods used in the collection of passenger flow information in our country.This method has high cost and can not reflect the long-term changing characteristics of bus passenger flow.With the continuous development and improvement of bus IC card technology,using bus IC card data acquisition method to collect passenger flow information has gradually become the mainstream information acquisition method.Bus IC card data acquisition method is simple,the data obtained contains comprehensive information,low cost,and without time constraints.It can provide data support for the study of the long-term characteristics of bus passenger flow.Based on the data of Qingdao bus IC card,this paper starts with the original data of bus IC card,and studies the methods of bus IC card data preprocessing,bus trip characteristics analysis and bus short-term passenger flow prediction.The specific research contents are as follows:Firstly,the data structure and preprocessing method of bus IC card are studied.Based on the data of Qingdao Bus IC Card,the data structure is analyzed and the data preprocessing method of Bus IC Card is discussed.Secondly,the characteristics of bus passenger flow are studied.Based on the data of IC card in Qingdao city in one week,the data are analyzed from two aspects: time characteristics of bus passenger flow and travel characteristics of different groups,including one-day passenger flow characteristics(working days and weekends),time-division passenger flow characteristics(morning peak,off-peak,evening peak),travel characteristics of elderly card group and student card group and ordinary card group.Then,the short-term bus passenger flow forecasting method is studied.Combining the dynamic change and non-linear characteristics of bus passenger flow data,ARIMA model,namely differential autoregressive moving average model and NARX neural network model,namely non-linear autoregressive neural network model,are selected to study.On the basis of considering the shortcomings of the above two methods,the NARX neural network model is improved by genetic algorithm.The GA-NARX neural network model is obtained,and the model used by the Institute of Bus Short-term Passenger Flow Forecasting is determined.Finally,a case study and a comparative evaluation are carried out.This paper chooses oneweek bus IC card data of Qingdao City,establishes ARIMA model,NARX neural network model and GA-NARX neural network model based on genetic algorithm to predict short-term bus passenger flow,and analyses and compares the prediction accuracy and error of the three models.The results show that the three models are based on legacy.The improved NARX neural network model has the highest accuracy in short-term passenger flow forecasting.