BRT Vehicle Travel Time Prediction Based on SVM and Kalman Filter

Abstract

An integrated BRT vehicle travel-time prediction model was proposed that used the support vector machine (SVM) to predict the initial travel time and applied the Kalman filter algorithm to dynamically adjust the results of the predicted travel time. Based on the GPS data, a case study of the BRT line 2 in Chaoyang district, Beijing, was conducted with the help of the proposed prediction model. BRT vehicle travel time during the morning peak hour and the off-peak hour was predicted by both the proposed model and the Kalman filter model. The results prove that the proposed model is more suitable for predicting the BRT vehicle travel time with a high prediction accuracy, and the accuracy for the off-peak hour is higher than the one for the peak hours.