A simulation-based mult-objective genetic algorithm (SMOGA) for transportation network design problem

In the conventional transportation network design problem, travel demand is assumed to be known exactly in the future. However, there is no guarantee that the travel demand forecast would be precisely materialized under uncertainty. This is because travel demand forecast is affected by many factors such as economic growth, land use pattern, socioeconomic characteristics, etc. All these factors cannot be measured accurately, but can only be roughly estimated. Another issue in many existing transportation network design problems considers only one objective or a composite objective with a priori weights. It may be more realistic to explicitly consider multiple objectives in the transportation network design problem. We incorporate both travel demand uncertainty and multiple objectives into the transportation network design problem. It is formulated as a stochastic bi-level programming problem (SBLPP) where the upper level represents the traffic manager and the lower level represents the road users. To solve this SBLPP, a simulation-based multiobjective genetic algorithm (SMOGA) is developed. Numerical results are provided to demonstrate the feasibility of SMOGA