A Kriging-based optimization method for meeting point locations to enhance flex-route transit services

As a promising on-demand transportation mode in low-demand areas, flex-route transit, has attracted much attention in the transportation research field. However, unexpectedly high demand levels caused by travel uncertainty impact the reliability and development of flex-route transit services. Although the meeting point strategy can deal with this problem effectively, selecting a location for the meeting points can substantially influence the performance of this strategy. In this study, meeting point location selection is modeled as a simulation-based optimization (SO) problem, and a Kriging-based global optimization method using a Pareto-based multipoint sampling strategy (KGO-PS) is proposed to solve this problem. Through comparison of several typical benchmark functions with other counterparts, the effectiveness of KGO-PS has been verified. Moreover, a real-life flex-route transit service is employed to construct the SO problem, and the optimization results show that the proposed algorithm can improve the performance of flex-route transit services under unexpectedly high demand levels.

Automated summary

Flex-route transit, as a promising transportation mode in low-demand areas, has received considerable attention in the transportation research field. However, the reliability and development of flex-route transit services are affected by unexpectedly high demand levels caused by travel uncertainty. This study proposes a Kriging-based global optimization method using a Pareto-based multipoint sampling strategy to solve the problem of meeting point location selection, which substantially influences the performance of the meeting point strategy. The optimization results using a real-life flex-route transit service show that the proposed algorithm can improve the performance of flex-route transit services under unexpectedly high demand levels.