Research On Prediction Technology Of Bus Arrival Time And Construction Of Dynamic Transfer System

With the wide application of Internet of things,big data and other information technologies in the transportation industry,the traditional transportation passenger transport began to transform and upgrade to intelligent transportation,so as to provide better ride experience for passengers.Among them,providing accurate public transport dynamic transfer service(DZSJ)to passengers is an important link to realize urban intelligent transportation,and thus puts forward an urgent demand for accurate prediction of bus arrival time.However,due to the complexity of urban road traffic environment,it is very difficult to accurately predict the arrival time of public transport,which has become the key to affect the dynamic transfer service.Therefore,this paper studies two kinds of bus arrival prediction models to meet the demand of dynamic transfer service for arrival time prediction under different needs of passengers.Prediction model of bus arrival time based on BP neural network and real-time bus location data(DTHC),which integrates BP neural network and real-time bus location data,can estimate the time required for passengers from departure to destination for a certain time point;The bus arrival time prediction model(RHMX)based on the fusion model can dynamically predict the arrival time according to the input data.For DTHC model,when the real-time location data of bus is complete,the real-time data is directly used to predict the bus arrival time;when the real-time location data of bus is missing,the BP neural network model is used to predict the bus arrival time,and the correction layer is introduced to improve the prediction accuracy.For the RHMX model,two sub models are constructed,one is the prediction of stop time in the station and the other is the prediction of running time between stations.The former uses the iterative method to predict the stop time of the running buses in each downstream station,and the latter also uses the iterative method to predict the running time of the running buses in each downstream road section(taking the station as the dividing point).Through these two models,a group of time series data of running time and stopping time between stations can be predicted.Then,LSTM is used to fuse the time series data from the two sub models to get the bus arrival time.Finally,Kalman filter is used to dynamically adjust the prediction result of LSTM to avoid the influence of abnormal data on the prediction value,so as to obtain more accurate arrival time of public transport.A dynamic transfer system for public transport is constructed to test and verify the validity and accuracy of the bus arrival time prediction method.Select the data of No.10 bus line in Guilin to verify the prediction method of this paper.DTHC is 3% lower than the best performing linear regression(LR)in root mean square error(RMSE)and 4%lower in mean absolute error(MAE);RHMX is 6% higher in RMSE and 7% lower in MAE than linear stack;RHMX is 9% lower than DTHC in MAE and 12% lower in RMSE.The prediction model proposed in this paper can meet the needs of practical application in accuracy and stability..