具有学习效应的半导体晶圆制造绿色车间调度问题研究

Research on Green Job Shop Scheduling Problem of Semiconductor Wafers Manufacturing with Learning Effect

针对最小化最大完工时间、总碳排放以及总拖期时间的具有学习效应的半导体晶圆制造绿色车间调度问题,构建了双影响因素的新型学习效应模型,提出了改进的多元宇宙优化算法,并对其收敛性进行证明。通过对初始种群进行反向学习、宇宙个体进行莱维飞行扰动和对外部档案中的个体进行邻域搜索变异更新,产生新的父代个体,扩大了种群的多样性,避免算法陷入局部最优。通过对小规模和大规模测试算例的仿真实验,以及利用改进算法求解具有异质性机器的学习型半导体晶圆制造绿色车间调度问题,验证了本文所提出的算法对于求解具有学习效应的半导体晶圆制造绿色车间调度问题的有效性和可行性。

In order to minimize makespan,total carbon emissions and total tardiness,we construct a new learning effect model with double influencing factors firstly.Thus an improved Multi-Verse Optimizer algorithm is proposed and its convergence is also proved.By performing reverse learning on the initial population,perturbing the individuals in the universe population by Levy flight,and mutating and updating the individuals in the external archive by neighborhood search,new parent individuals are generated,which expand the diversity of the population and avoid the algorithm falling into local optimum.Through the simulation experiments of small-scale and large-scale test cases,and the use of improved algorithm to solve the green shop scheduling problem of learning semiconductor wafers fabrication with heterogeneous machines,it is verified that the proposed algorithm for solving the green shop scheduling problem of semiconductor wafers fabrication with learning effect is valid and feasible.