RFID E-Plate Data based Bus Travel Time Prediction Considering Traffic Flow Diversion Rate

The conventional bus travel time prediction (bus-TTP) generally utilizes the Global Position System (GPS), video detection, etc. However, they are vulnerable to ambient, weather, illuminance, electromagnetic interferences or other additional factors. To improve the performance of TTP, we study the Radio Frequency Identification (RFID) electronic license plate (E-Plate) data based bus-TTP and proposed an improved Kalman filter (i.e., the Adaptive Extended Kalman Filter, AEKF) algorithm considering traffic flow diversion rate (TFDR). Firstly, the RFID readers mounted on road gantry read the critical information from the electronic tag deployed on vehicles. Secondly, the received traffic information is modulated and demodulated as the available information. Finally, the proposed bus-TTP algorithm is used to predict bus travel time, which makes the predicted bus travel time better approach the actual value. The experimental results show that the AEKF algorithm Considering TFDR improves the dynamic performance of the filtering and better inhibits the filtering divergence process in the case of bus travel time changing abruptly during rush hours.