Multi-objective optimization of real-time customized bus routes based on two-stage method

Customized public transit, as a diversified passenger service mode proposed under the new situation, not only deals with the comfortable and convenient commuting needs of most commuters, but also improves the efficiency of road use effectively and alleviates the road traffic pressure in the morning and evening rush hours. Currently, various routing optimization methods based on static passenger requests have been applied to the optimization design of customized bus routing in major cities, but the static passenger requests cannot satisfy the dynamic passenger requests of random users at present Therefore, it is urgent and necessary to investigate real-time customized bus routing optimization methods based on static passenger requests and dynamic passenger requests. In this paper, our goal is to present an approach to tackle two important issues of real-time customized bus route optimization under stochastic user demand: one is that how to deal with dynamic travel requests in a specific road network to improve passengers' satisfaction with customized bus services; the other is that how to employ specific methods to achieve real-time updating of customized bus routes to meet the random user's travel requests. Starting from the whole operation process of customized bus, we firstly construct a multi-objective optimization model of customized bus route, and clearly depicts the four processes of customized bus operation in the form of mathematical expressions. Second, we design a two-stage method based on NSGA-II algorithm to solve the model, in which the first stage deals with static bus travel requests to obtain the initial customized bus route optimization scheme, and the second stage deals with dynamic bus travel requests to update the customized bus route in real time. Finally, a two-stage case is illustrated to verify the correctness and validity of the model and the method for solving the real-time customized bus route optimization problem under stochastic user demand. (C) 2019 Elsevier B.V. All rights reserved.