Optimization of adaptive dispatch strategies for on-demand feeder transit considering uncertain demand
On-demand feeder transit(ODFT)is an innovative example of traditional transit systems integrating e-hailing technologies.It is normally utilized to connect a transport hub(e.g.,metro sta-tion or airport)with flexible-route door-to-door services for passengers spread across a distant region.However,ODFT's vehicle dispatch strategies still mainly follow traditional methods based on sched-ulesor frequencies.In cases of uncertain passenger arrivals,these dispatch strategies may lead to un-expected oversaturation and undersaturation.To address these issues,the paper proposes adaptive dis-patch strategies to dynamically determine the dispatches based on actual passenger/request arrivals.It introduces the concept of occupancy target to build dispatch models using a fixed dispatch thresh-old and a time-varying dispatch threshold.The former sets a fixed occupancy target,while the latter utilizes a time-declining function.The proposed models are optimized to minimize the expected sys-tem cost,and a grid search method is employed to solve the optimization problems.Numerical experi-ments show that time-varying threshold models outperform counterparts(e.g.,headway-based and fixed threshold models)when the demand mean is low and variation is large.The costs saving can reach up to 18%.Sensitivity analysis shows that the advantage of these models is also influenced by passengers'value of time.Adaptive dispatch models have broad application potential and can be ap-plied,with slight modification,to other transit systems.
万方数据
维普
