Study On Prediction Of Bus Dwell Time At The Station Based On GA-BP Neural Network

Bus stops are places where buses can stop to serve passengers to get on,get off and transfer.Bus operation lines are composed of bus stops and intersection sections.The operation time of the whole bus line is generally composed of bus dwell time,section travel time and intersection delay time.Therefore,accurate prediction of bus dwell time at the station is the key link to predict bus running time,implement bus priority signal control and build intelligent bus system.The accurate prediction of bus dwell time can improve the accuracy of bus run time prediction,improve the accuracy and reliability of information service at bus stop,guarantee the good operation of intelligent bus dispatching system,thus improve the service level of public transport system and increase the attraction rate of bus trip mode.This paper divides the process of bus arrival and stop into three stages:deceleration,parking service and acceleration,and analyses the time composition and influencing factors of bus arrival and stop.GPS technology is used to collect the infornation of bus operation and parking.Aiming at the shortcomings of GPS original data,a method of data preprocessing is proposed to obtain the data of bus entry and exit,and then an algorithm for obtaining the time of bus stop is proposed and tested.Through case analysis,it can be seen that the errors between bus stop time and actual value obtained from GPS data are within the allowable error range,and can be used as data source of bus dwell time at the station prediction method.According to the bus dwell time acquired from GPS data,the general distribution characteristics of bus dwell time at the bus station are analyzed and the distribution fitting function is given.The historical dwell time and real-time data of bus are selected as input variables of the prediction model of bus dwell time at bus stops.Finally,the prediction models of bus dwell time based on BP neural network and GA-BP neural network are established respectively.The key structure,learning process and evaluation method of the model are analyzed.Based on the actual bus dwell time,the prediction model of bus dwell time built in this paper is simulated and validated by using MATLAB software.The simulation results show that compared with the traditional BP neural network,the GA-BP neural network algorithm has higher prediction accuracy,stronger stability and reliability,and can more accurately predict the bus dwell time at the station.