基于惩罚和修复混合策略的约束处理方法

约束处理是约束优化的关键问题，特别是非线性约束的处理一直缺少特别有效的解决方法，将惩罚函数法与修复策略结合使用，可以有效地避免迭代过程中大量非可行解的产生，使得约束优化问题在惩罚函数和修复算子的协同作用下收敛于全局最优，较好地解决了在遗传算法约束优化问题中单独使用惩罚和修复方法时一些难以解决的问题。基于随机方向法构造的修复算子作用效果显著，采用多个测试函数对算法进行检验，均能较好地收敛于可行域中的最优解，验证了算法的可靠性。

Technique of constrain processing based on punishment and reparation strategy

Constrain processing is the most difficult problem in constrained optimization, especially for nonlinear constrain. Until now there is no very effective technique for this kind of questions. The punish function and the reparation strategy are combined for the application of genetic algorithms in nonlinear constrained optimization, the constrained optimization genetic algorithms could converge more quickly. This method could avoid producing unfeasible solutions, and solve the problem of genetic algorithms application in nonlinear constrained optimization. Many testing functions are made by test some methods. And the results show that they could all converge in feasible area excellently.