多目标遗传算法适应度函数权重设置的改进

Improvement of Moderate Weight to Fitness Function of Multi-Objective Genetic Algorithm

针对传统方法中权值分配不合理会导致某个目标下非支配解遗失的问题,提出了均值自适应法、均值波动自适应法以及均值调节与传统算式相结合的三种的方法,使遗传算法中权值系数的设置得到了很大程度的改善,一定程度上提高了多目标下遗传算法的优化性能,并通过在车桥厂实例中的应用,取得了良好的效果,证明了此种方法的可行性和有效性。

The immoderate weights of fixed weight MOGA could lead to missing nondominated solutions about one of objectives. To solve this problem, three new methods are proposed： average adaptive weight approach, average and fluctuating adaptive evaluation function approach, average leading weight approach. They all greatly improve the performance of genetic algorithm. With the implement of these methods, they are proved to be feasible and efficienct.