ESTIMATING TRAIN SPEEDS FOR TRAIN PREEMPTION USING MULTIPLE SENSOR DATA

摘要

The ability of second generation train detection technologies to accurately measure train speed is evaluated. Train speed data were collected using video detection and Doppler radar at two locations in Nebraska, U.S. Then, these train speed measurements were fused using a discrete Kalman filter model and the speed data from Doppler radar, video detection and Kalman filter were compared using two MOEs: RMSE and MAPE. The results show that both video and radar provide accurate train speed measurements and that a Kalman filter can reduce the noise in the speed measurements over that found in either sensor.