Abstract: Aiming at solving the hybrid flow shop scheduling problems, the standard DE algorithm had the disadvantages of easily falling into local extremum. Therefore, an improved DE algorithm was proposed to solve the simulation optimization model based on the minimization of makespan. The proposed algorithm was combined with the reverse learning strategy to generate the initial population, the  adaptive difference factor was further introduced into DE, and the Metropolis criterion of simulated annealing algorithm was introduced in the individual selection mechanism, which effectively improved the global search ability of the algorithm. Finally, the simulation results of the proposed algorithm and the classical algorithms were compared based on different scale examples to verify the effectiveness and superiority of the proposed improved DE algorithm.

Key words: hybrid flow-shop, differential evolutionary(DE) algorithm, reverse learning strategy, Metropolis criterion, makespan

摘要： 对于求解混合流水车间调度问题，标准差分进化算法存在易陷入局部极值的缺点，为此，以最小化最大完工时间为目标函数建立了仿真优化模型，并提出了一种改进差分进化算法进行求解。将所提算法结合反向学习策略生成初始种群，在差分进化中进一步引入自适应差分因子，并在个体选择机制中引入模拟退火算法的Metropolis准则，有效提高了该算法的全局搜索能力。最后基于不同规模算例对所提算法和经典算法进行了仿真实验结果对比，验证了所提改进差分进化算法的有效性和优越性。

关键词: 混合流水车间, 差分进化算法, 反向学习策略, Metropolis准则, 最大完工时间