Abstract: The global solutions of the constrained optimization problems often locate on or are near the boundary of the feasible region, and it is hard to get the global solutions. A novel hybrid evolutionary algorithm based on Davies, Swann and Campey search method（D.S.C.）, referred as Improved D.S.C.（I.D.S.C.） is proposed. Starting from a random solution, a solution on or near the boundary of the feasible region is got by D.S.C.search method with the revised stop criterion. Then, a new fitness function is used. It can automatically select potential high quality solutions without discussion of the different situations as the similar fitness did; meanwhile, avoid throwing away potential high quality solutions, the strategy of keeping a certain number of feasible solutions is adopted. At last, the computer simulations are made on 5 benchmark problems and the results demonstrate that the proposed algorithm is effective.

Key words: constrained optimization, D.S.C.method, evolutionary algorithm

摘要： 约束优化问题最优解通常分布在可行域边界上或在可行域边界附近，对其求解比较困难。对此类问题提出了一种基于D.S.C.（Davies，Swann，Campey）法的混合进化算法，简记为I.D.S.C。从某个随机解出发，利用改进了停止法则的D.S.C.法进行一维搜索，使得搜索到的解在可行域边界或其附近；采用了新的适应度函数，可以自动选择有潜力的解，无需像类似方法分情况进行选择；同时，为了避免丢掉好的解，算法还启用了保留一定数目可行解的策略。对5个标准的测试函数进行了实验，结果验证了算法的有效性。

关键词: 约束优化, D.S.C.法, 进化算法