Flex-route demand response transit scheduling based on station optimization

Abstract

In order to increase the proportion of passengers served by the flex-route demand response transit (DRT) and reduce its service cost, we propose an improved D-<italic>k</italic>-means clustering algorithm that based on DBSCAN clustering algorithm and <italic>k</italic>-means clustering algorithm, which could determine fixed stations and alternative stations according to passengers' needs. Besides, we design a flex-route demand response transit service system and establish the flex-route demand response transit scheduling model which is solved by an accurate algorithm. Finally, the feasibility of the scheduling model is verified by the example in the southern part of Jiexi County of Guangdong Province in China. The results indicate that the model serves 95.3% of passengers, and comparing with regional flexible bus, the operating cost and time are reduced by 9% and 5%, respectively. The algorithm is fast and stable.