多群体混合进化算法求解IPPS问题  

Multi-group hybrid evolutionary algorithm for solving IPPS problem

作　　者：杜轩[1,2] 潘志成[1] 张屹[1,2] 陆曈曈[3] Du Xuan Pan Zhicheng Zhang Yi Lu Tongtong(College of Mechanical ＆ Power Engineering Hubei Key Laboratory of Hydroelectric Machinery Design ＆ Maintenance College of Economy ＆ Management, China Three Gorges University, Yichang Hubei 443002, China)

机构地区：[1]三峡大学机械与动力学院,湖北宜昌443002 [2]三峡大学水电机械设备设计与维护湖北省重点实验室,湖北宜昌443002 [3]三峡大学经济与管理学院,湖北宜昌443002

出　　处：《计算机应用研究》2017年第9期2594-2598,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(51275274;71501110);湖北省自然科学基金资助项目(2014CFC1141)

摘　　要：针对工艺规划与调度集成(integration of process planning and scheduling,IPPS)问题求解复杂性,为提高求解效率,设计了包含探索种群、寻优种群和最优种群的多群体混合进化算法,通过运用混合遗传算法和基于聚类淘汰机制的差分进化算法分别更新探索种群中工艺链和加工顺序链,保持可行解多样性和差异性;然后利用克隆领域搜索算法完成寻优种群中可行解的克隆和领域搜索,进一步提高种群质量;最后按照精英保留策略更新最优种群获得全局最优解。通过实例计算对比,结果显示算法搜索效率和求解质量均有明显改善,且稳定性较好,表明该算法求解IPPS问题的可行性及优越性。For integrated process planning and scheduling （IPPS） problems solving complexity, in order to improve computa- tional efficiency, this paper designed a multi-group hybrid evolutionary algorithm including explore population, optimum popu- lation and optimal population. The algorithm used hybrid genetic algorithm and the differential evolution algorithm based clus- tering mechanism to update processing chains and process order chains of explore population respectively, kept the diversity and difference of feasible solutions. Then it used the clone and field searching algorithm to complete clone and search nf feasi-ble solutions in the optimum populations, improved the quality of populations further. Finally, through the example calculation and comparison, the calculation results indicate that the algorithm can improve searching efficiency and solving quality, and has good stability, which shows the feasibility and superiority of the algorithm to solve the IPPS problem.

关 键 词：工艺规划与调度 聚类 差分进化算法 克隆 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]