Model and Algorithm of Multi-objective Discrete Transportation Network Design under Stochastic Demand among OD Pairs

In order to solve the multi-objective optimization problem under uncertain traffic demand among practical OD pairs, a bi-level programming model was proposed to optimize the traffic management, environment protection, investment cost, and user behavior for stochastic multi-objective discrete transportation network design. The upper-level programming model was constructed using the chance constrained model and the ideal point model for multi-objective optimization, and the lower-level programming model was constructed using the user equilibrium assignment model under a fixed traffic demand. To ensure the solution accuracy of the proposed model, a genetic algorithm based on Frank-Wolfe algorithm, Monte-Carlo simulation, and adaptive niche technology was designed, and its corresponding program was developed using Matlab. In addition, the model and algorithm were tested in the Nguyen-Dupuis network. The result indicates that the model can reflect the objectives and constrains of practical network planning, and the algorithm is global convergent, hence providing a reference for the practical transportation planning.