Section: New Results

MO-DYNAMOP

Participants: S. Jacquin, L. Jourdan, E-G. Talbi

The proposed method, MO-DYNAMOP generalize to multi-objective optimization, DYNAMOP, a state of the art optimizer, which was successfully applied to several MO problems. The specificity of this method is to combine aMO dynamic programming (MO-DP) with a MO evolutionary algorithm (MOEA). MO-DYNAMOP is applied to the first stage of the MO-UCP problem including minimization of gas emission. Since the second stage of the problem is now multi-objective, each solution of the first stage problem induces an entire Pareto front of the second stage problem. MO-UCP is solved by assigning an approximation of this Pareto front to each solution of the first stage problem. A comparison study with methods previously proposed in literature is performed. Experiments indicate that MO-DYNAMOP performs considerably better.