Study Of Dynamic Bus Arrival Time Prediction

To provide the passengers with the bus travel time and arrival time information can dramatically improve the level of service of traditional bus systems.The factors affecting bus travel time are complicated and the prediction is difficult,but the passengers have high requirements for accurate and real-time bus arrival time.It requires to predict the bus arrival time dynamically in order to help passengers better plan their trips,minimize waiting times at bus stops and reduce passengers’ anxiety.At first different bus arrival time prediction models are reviewed in this study.Then dynamic bus travel time prediction models based on support vector machines(SVMs)and Kalman filtering algorithm or artificial neural network(ANN)and Kalman filtering algorithm are proposed.In the proposed models,the well-trained SVM or ANN model predicts the baseline bus travel times from the historical bus trip data;the Kalman filtering dynamic algorithm can adjust bus travel times with the estimated baseline travel times from SVM or ANN model.To validate the effectiveness and feasibility of the proposed dynamic models,the performance of the proposed models is validated with the real-world data in Shenzhen,China.The results of dynamic models are compared with those of the pure Kalman model,pure SVM model and pure ANN model.The results show that the proposed dynamic models have better prediction performance in terms of prediction accuracy with good time efficiency.In order to further improve the prediction accuracy of dynamic bus arrival time prediction models,the common phenomenon,which addresses the cases that several bus routes share the same road segments and bus stops in the transit-oriented cities,is considered in this study.The variable of weighted average bus travel time of preceding buses of any routes is included in the prediction models.With the real-world data in Shenzhen,China,the results show that the models based on the multiple bus routes can further improve the prediction accuracy.