改进COOT算法求解多目标柔性车间调度问题  

Improved COOT Algorithm to Solve Multi-Objective Flexible Jobshop Scheduling Problem

作　　者：凌方平 吉卫喜 LING Fangping;JI Weixi(School of Mechanical Engineering, Jiangnan University, Wuxi, Jiangsu 214122, China)

机构地区：[1]江南大学 机械工程学院,江苏 无锡 214122

出　　处：《计算机工程与应用》2023年第22期307-314,共8页Computer Engineering and Applications

基　　金：山东省重大科技创新工程基金项目(2019JZZY020111)。

摘　　要：针对柔性车间调度的多目标优化问题,建立了以完工时间、机器总负荷、能耗为优化目标的模型,并提出了一种结合模拟退火的多目标COOT算法(multi-objective COOT algorithm combined with simulated annealing,MOCOOT-SA)进行求解。该算法通过引入存档集和Pareto解的理念,将原有的单目标COOT算法优化成多目标算法,并为其中特定个体选择新的邻域结构和更新方式,再融合模拟退火算法(simulated annealing,SA)优化局部搜索能力和收敛速度。最后选用合适的编解码方式,用MOCOOT-SA算法测试改进的基准算例,并与NSGA-Ⅱ算法、MOPSO算法的结果进行对比,得到各目标上的平均值优化比为0.013~0.047,最优值优化比为0.016~0.045。结果表明,该算法的优点是能更好地解决多目标柔性车间调度问题。Aiming at the multi-objective flexible jobshop scheduling problem(FJSP),a model with the completion time,total machine load and energy consumption as the optimization objectives is established,and the multi-objective COOT algorithm combined with simulated annealing(MOCOOT-SA)is proposed to solve it.The algorithm optimizes the original single-objective COOT algorithm into a multi-objective algorithm by introducing the concept of archive set and Pareto solution,and selects a new neighborhood structure and update method for specific individuals,and then integrates the simulated annealing algorithm(SA)to optimize local search ability and convergence speed.Finally,the appropriate codec method is selected,the MOCOOT-SA algorithm is used to test the improved benchmark example,and compared with the results of the NSGA-II algorithm and the MOPSO algorithm,the average optimization ratio of each target is 0.013~0.047,the optimal value optimization ratio is 0.016~0.045.The results show the advantage of this algorithm that it can better solve the multi-objective FJSP.

关 键 词：柔性车间 生产调度 多目标优化 MOCOOT-SA算法 

分 类 号：TH165[机械工程—机械制造及自动化] TP18[自动化与计算机技术—控制理论与控制工程]