Dynamic Dwell Time Determination and Optimization based on Motion Detection Algorithm

Dwell time, as an important parameter in timetabling, is also a significant temporal component in rail transit operation. In general, the dwell time is a fixed value, so that passengers have to wait on board before the doors close. Therefore, dynamic dwell time, which adjust itself according to the actual passenger boarding and alighting activities, is proposed in this paper. In other words, the train would close the door and depart when these passengers' activities are recognized actual finish. Acquisition device herein is used to collect the video date of passenger activities around the door area. The time point to close the doors can be determined according to the change of detected motion value (DMV) in video through dynamic motion detection algorithm using inter-frame subtraction method. This algorithm was conducted to one metro line in Beijing partly with 5 stations. The results show that the time points, determined by dynamic motion detection algorithm, better correspond to the actual passenger boarding and alighting activities. This study contributes a unique method to dynamic adjust dwell time in real operation with the support of video technology. It shortens passenger travel time, improves the quality of passenger service and increase the through capacity for operation.