Joint Optimization Of Transfer-based Customized Bus Transit Network And Passenger Assignment Under Elastic Demand

Due to the rough development of conventional public transportation,it is difficult to satisfy travelers’ needs for accessibility,efficiency and service quality in the expanding urban area,the development of diversified high-quality public transportation has become an inevitable trend in urban transportation development.As a flexible and intensive public transportation system,customized bus(CB)service is a vitality for urban public transportation system,where multiple travelers can share the seats if their itineraries are somewhat “adjacent” in a spatial-temporal sense.Customized bus is a flexible and on-demand bus service,which is designed to attract private car owners by providing high-quality service comparable to that of a car and at a much lower cost than private car.Therefore,a well-designed CB system has the potential to reduce the occupancy rate of roads and mitigate environmental pollution and traffic congestion.However,in view of the problem of high operational cost and relatively low operational efficiency of CB system resulting from the lack of consideration of elastic demand as well as the pursuit of directness when designing the CB transit roues.This paper take account of the elastic demand to capture the interaction between passenger demand and the customized bus transit network,the objective is to maximize the competitiveness of CB compared with private car under certain budget constraints,as well as achieving the Stackelberg equilibrium between the profit and the average passenger travel cost.Compared with the conventional CB transit network optimization methodology,this paper differs in the following three aspects: firstly,in terms of transit network structure,for the dispersed passenger travel demand,transfer is incorporated into the network structure to reduce the length of routes and detours,thus reducing passenger travel costs and achieving higher accessibility with fewer routes;secondly,in terms of elastic passenger demand,a bi-level model is applied to capture the interaction between passenger demand and transit network structure.Finally,in terms of transit assignment,this paper jointly optimizes the CB network structure and passenger-route assignment with limited capacity and fleet size,such that the vehicle capacity could be fully utilized and thus meet more passenger demand at lower operational cost.To solve the bi-level optimization model proposed in this paper,we propose a heuristic initial solution generation method,four types of deconstruction operators,reconstruction operators,and an adaptive learning mechanism based on simulated annealing algorithm.Then,the validity of the model and algorithm is verified using Mandl’s Swiss dataset.Finally,the capability of the proposed algorithm to handle large-scale real-world cases is verified by applying the algorithm to the customized buses passenger demand data in Tianhe District,Guangzhou City.The results show that the optimized scheme can significantly reduce the operating mileage(29.07%) and the average passenger travel time(22.3%) compared with the original scheme,and achieve higher transit efficiency with fewer customized bus resources by jointly optimizing the transfer-based transit network and transit assignment.