Abstract:

To improve the efficiency of the Ranking-Method in GAME, two handles are applied. The first handle is applied to reduce the computational time of GAME, which is based on the idea of "Divide-and-Conquer". Then another handle is applied to form the fuzzy ranking, which is meaningful in multi- objective problems with multi-constraints. In this handle, infeasible elitists are reserved and selected as parents for the next generation. These infeasible solutions do not match the constraints but they are closer to the real optimal results, they are reserved as a "bridge" to reach the true Pareto Front. Thus the new ranking method has much possibility to reach the true Pareto Front. In the end, two experiments, random number comparison and test problem OSY, are used for analyzing and comparing the e±ciency of the new and original method.