如何有效地均衡可行区域与不可行区域的搜索是约束优化中的关键问题。为使进化算法获得可行的全局最优解,分析了在进化过程中如何对待好的不可行解的问题,通过分析随机排序中比较概率对可行解最终位置的影响,提出一种动态随机选择策略,并以多个体差分进化为框架实现了相应算法。实验对比分析结果说明了这一策略的有效性。
It is a key problem to balance searching for feasible and infeasible areas efficiently. In order to effectively locate the feasible global optimum of evolution algorithm, this paper analyzes how to treat the promising infeasible solutions investigated. Through analyzing the influence of the comparison probability in stochastic ranking on the final position of the feasible solution, a novel dynam ic stochastic selection strategy is proposed, and related algorithm implementation within the framework of muhimember differential evolution is also discussed. Experimental results on common bench mark functions demonstrate the effectiveness of the strategy.