1.Faculty of mechanical and electrical engineering, Kunming University of Science and Technology;2.Faculty of Information Engineering and Automation, Kunming University of Science and Technology;3.Institute of Control and Decision, Tsinghua University
针对低碳分布式装配置换流水车间调度问题 (LC_DAPFSP), 建立以同时最小化总能耗和总完工时间为优化目标的问题模型, 进而提出一种多维分布估计算法 (MEDA) 进行求解. 首先, 采用随机方法和启发式算法共同生成初始化种群; 其次, 建立基于矩阵立方体的概率模型, 用于合理学习并积累优质解的块结构信息和序关系信息, 同时设计有效采样机制对概率模型采样以生成新种群, 从而可合理引导算法搜索方向并发现解空间中的优质解区域. 然后, 为平衡算法的全局探索与局部开发能力, 提出基于问题特性的变邻域局部搜索方法, 可对全局搜索发现的优质解区域进行细致搜索. 最后, 仿真实验与算法对比验证MEDA是求解LC_DAPFSP的有效算法.
For the low carbon distributed assembly permutation flow-shop scheduling problem (LC_DAPFSP), a problem model with the optimization goal of simultaneously minimizing the total energy consumption and the makespan is established, and a multidimensional estimation of distribution algorithm (MEDA) is proposed to solve it. Firstly, a population is initialized by hybriding a random method and a heuristic algorithm. Secondly, a matrix-cube-based probabilistic model is developed to reasonably learn and accumulate the information of the job blocks and the jobs’ order from the superior solutions, and an effective sampling mechanism is designed to sample the probability model to generate new population, so as to reasonably guide the searching directions and find the promising regions in solution space. Then, to balance the exploration and the exploitation capabilities of the algorithm, a problem-dependent variable neighborhood search method is developed to perform an in-depth exploitation in the promising regions found by the global search. Finally, Simulations and comparisons demonstrate that the proposed MEDA can effectively solve the LC_DAPFSP.