Research On Passenger Flow Control Policy And Bus Bridging Service Optimization Under Uncertainty In Urban Rail Transit

Passenger flow oversaturation during peak hours is a key challenge facing metro systems worldwide.It affects the service level and operational efficiency as well as results in safety concerns induced by passenger accumulation.This thesis develops related management strategies from demand and supply aspects in order to alleviate oversaturations.The passenger flow control and bus bridging service design models under uncertainty situations are developed.Meanwhile,the solution algorithms based on online and robust optimization theory are devised to solve the proposed models efficiently.The main contents of this thesis include the following four aspects.(1)This thesis studies the theoretical model and solution algorithm for online passenger flow control policy in the single train setting.Consider the stochastic passengers origin-destination(OD)demand information reveals only when the train arrives at its origin station,this thesis develops stochastic dynamic programming(DP)with fairness constraints.The DP model aims to maximize the number of boarding passengers as well as ensure certain service level to each OD pair.This thesis translates the DP model into a multi-objective stochastic DP problem by setting the desired service level target for each OD pair.Meanwhile,this thesis also characterizes the necessary and sufficient conditions for certain service level targets to be feasible under any capacity level.An online flow control policy based on the approximate DP is designed,which can sequentially determine flow control decisions at each station after knowing the realized demands.The numerical experiments using Beijing transit data reveals that the devised policy outperforms the first-come-first-serve policy in terms of efficiency and fairness.(2)This thesis analyzes the online collaborative passenger flow control problem in the multiple train setting.Note that passenger demands faced by each train include both the new arrival demands and passengers stranded by previous trains.This thesis develops a stochastic DP with fairness constraint,to optimize the number of boarding passengers during peak hours and ensure certain service level to each OD pair.This thesis exploits the solution framework in the single train setting to design a collaborative passenger flow control policy,which can sequentially determine the flow control decisions for each train at each station.The case study with Beijing transit data reveals that the devised policy in this research can improve the metro system efficiency as well as ensure service level fairness among different OD pairs.(3)This thesis studies bus bridging network design under normal rail transit conditions.A two-step model framework is developed to incorporate the case that bus travel time and passenger demand may deviate from their nominal values.In the first step,the column generation procedure is designed to identify the candidate set of bus transit lines and passenger paths.Then a stochastic linear programming model is developed in the second step,which aims to minimize the expected total system cost as well as satisfy the bus fleet and passenger demand constraint in expectation.An online primal-dual algorithm based on the online convex optimization theory is designed to solve the proposed model.The numerical experiments using Beijing transit data reveal that the devised algorithm can obtain near-optimal solutions with high computational efficiency.(4)This thesis studies designing bus bridging services to evacuate stranded passengers under rail transit disruptions.First of all,a deterministic bus bridging service design model is formulated to minimize the total system cost.Then its robust counterpart is developed to incorporate the variation of the bus travel time in its uncertainty set.The robust counterpart can ensure the solution remains feasible under uncertain bus travel time and optimize the objective value in its worst case.The improved column generation procedure is devised to solve this problem efficiently.Numerical results with Beijing transit data reveal that the developed method can efficiently evacuate the stranded passengers and mitigate the adverse effects.