A novel multi-objective evolutionary algorithm for transit network design and frequency-setting problem considering passengers’ choice behaviors under station congestion

The transit network design and frequency-setting problem (TNDFSP) plays a critical role in urban transit system planning. Due to the conflict between the level of service and operating costs, extensive research has been conducted to obtain a set of trade-off solutions between the interests of users and operators. However, most studies ignored the effects of station congestion in TNDFSP, resulting in unrealistic solutions or a failure to achieve optimal design schemes. Therefore, this study investigates the multi-objective optimization of TNDFSP considering users’ choice behaviors under station congestion. To address the problem, a multi-objective bilevel optimization model is first formulated. The upper level is a bi-objective optimization model with two conflicting objectives: minimizing users’ cost and minimizing operator’s cost. The lower-level problem is a passenger assignment problem under station congestion. Moreover, a novel multi-objective evolutionary algorithm based on objective space decomposition (MOEA-OSD) is proposed to solve the complex problem. When dealing with multi-objective optimizations, a decomposition mechanism is developed to convert the problem into multiple subproblems. These subproblems are optimized using an evolutionary approach with newly designed selection process and elite preservation strategy to achieve desirable convergence and diversity. The computational results obtained using Mandl’s benchmark demonstrate the efficacy of MOEA-OSD and the advantage of the proposed model in achieving more comprehensive trade-off solutions.