应用重启改进进化算法提高柔性作业解质量

Improving Search Performance for Coevolutionary Algorithm with Restart Strategy

重启策略有效提高了概率算法性能。为将重启思想引入协同进化算法,改进了涉及多种群的协同进化算法收敛判断条件。以进化过程中相同解码值的重复解码次数衡量协同种群的稳定状态,用于结束或重启搜索过程。引入重启后的协同进化算法用以求解柔性作业调度问题。实验表明,改进后的算法能有效跳出局部最优,提高解质量和搜索效率。从而为应用协同进化算法求解其他组合问题时提高算法性能提供了一条可行有效的新途径。

Restart strategy improves performance of probabilistic algorithms.To introduce restart strategy to coevolutionary algorithms,a new method to evaluate status of populations was proposed and applied to terminate or restart search process.Improved coevolutionary algorithm was applied to dealing with complex flexible job-shop scheduling problem.Experimental results of extensive computational simulations show that the improved coevolutionary algorithm with restart strategy provides higher quality solutions and better search efficiency no matter the test-bed problem is large or not.Coevolutionary algorithms are widely applied in many fields and this improvement presents a feasible and effective way to enhance its application extension and depth.