集成多目标遗传算法在货位分配中的应用  

Ensemble Multi-Objective Genetic Algorithm with Application to Automated Warehouse Scheduling

作　　者：蔡安江[1] 蔡曜 郭师虹[2] 耿晨 CAI An-jiang;CAI Yao;GUO Shi-hong;GENG Chen(Xi’an University of Architecture and Technology,College of Electrical and Mechanical Engineering,Shanxi Xi’an,710055,China;Xi’an University of Architecture and Technology,College of Civil Engineering,Shanxi Xi’an 710055,China)

机构地区：[1]西安建筑科技大学机电工程学院,陕西西安710055 [2]西安建筑科技大学土木工程学院,陕西西安710055

出　　处：《机械设计与制造》2019年第5期95-98,共4页Machinery Design & Manufacture

基　　金：教育部"蓝火计划"产学研联合创新项目(2014-LHJH-HSZX-018)

摘　　要：根据效率优先原则、稳定性原则建立适合同端式出/入库立体仓库的多目标货位分配模型。基于向量评估、非支配排序、小生境Pareto等理论方法设计了三种多目标遗传算法(MGA)。根据集成学习理论,将若干多目标遗传算法集成,构建集成多目标遗传算法(EMGA),使优化算法适应搜索过程的任意阶段。以某铝厂实际工况进行仿真验证,结果表明,集成多目标遗传算法受问题规模影响小,收敛速度快,较单独其他多目标遗传算法性能更优越,是适用于立体仓库调度研究的高效算法。According to the principle of efficiency and stability,a multi-objective scheduling model which is appropriate for the same I/O station warehouse is presented. Three kinds of multi-objective genetic algorithms(MGA)based on the theory of vector evaluated,non-dominated sorting and niched Pareto are designed. Those MGAs are integrated on the basis of ensemble learning theory to create a ensemble multi-objective genetic algorithm(EMGA). And the optimization algorithms is suitable during different stages of the search process. The obtained results have shown that the EMGA was not appreciably affected by the scale of the problem,and the convergence rate is faster to other MGAs. The EMGA is an efficient algorithm which is suitable for the scheduling of the automated warehouse.

关 键 词：立体仓库 集成多目标遗传算法 货位分配 货位优化 

分 类 号：TH16[机械工程—机械制造及自动化] TP273[自动化与计算机技术—检测技术与自动化装置]