Multi-Population Evolutionary Algorithm for Solving Constrained Optimization Problems

Abstract

A novel multi-population evolutionary algorithm (MPEA) is presented, which can solve the constrained function optimization problems rather efficiently. The MPEA adopts three populations with different multi-parent crossover operators. So each population emphasizes particularly on different searching regions and the complementarity of these three crossover operators can enhances the diversity of individuals, which improves the search ability of the MPEA dramatically. And during the MPEA runs, the three populations exchange the best solution in each generation to adjust its search direction to the possible optimum solution. Experiments have been carried on several benchmark functions to test the performance of the presented MPEA. Numerical results show that MPEA is highly competitive with other algorithms in effectiveness and generality.