Abstract: Distributed manufacturing and scheduling system has become the main production mode in large agricultural machinery production enterprises under the background of economic globalization. In this paper, the distributed two-stage heterogeneous hybrid flow-shop scheduling problem (DHHFSP) is constructed based on the property of the agricultural machinery production. A knowledge-guided estimation of distribution algorithm (KEDA) is proposed to solve the DHHFSP. The DHHFSP contains three sub-problems, which are factory allocation, the sequence of job processing and arrangement processing machine. Multiple heuristic structure and random method are proposed for population initialization in KEDA. The candidate solution is optimized iteratively. The knowledge guided reinforcement mechanism and various local search strategies are proposed through analysis on the characteristics of DHHFSP problem. The utilization rate of key processing stages of processing resources can be improved. Through the simulation experiments, the proposed algorithm is compared with other three kinds of algorithms to verify the effectiveness and stability of the improved distribution estimation algorithm. The experimental results show that the efficiency of KEDA algorithm for DHHFSP problem can be improved by using the characteristics of DHHFSP problem to guide the evolution process of the algorithm.

Key words: estimation of distribution algorithm, hybrid flow-shop scheduling, distributed scheduling, knowledge-guided

摘要： 在经济全球化背景下，分布式制造和调度系统已成为大型农机生产企业的主流生产模式。针对农机生产过程中多品种小批量的生产特点，构建出一种分布式两阶段异构混合流水车间调度问题模型，提出了一种知识引导的分布估计算法，求解分布式异构混合流水车间调度问题模型的子问题：工厂分配、工件加工顺序和加工机器分配。改进的分布估计算法融合了多种启发式构造和随机方法进行种群初始化，并对候选解进行迭代优化，通过对求解问题的特性进行分析，提高关键加工阶段加工资源的利用率，对于不同规模的调度问题提出了相应的知识引导的强化机制和多种局部搜索策略。通过仿真实验，将提出的算法与其他三类算法进行对比，验证了改进的分布估计算法的有效性和稳定性。实验结果表明，利用调度问题特性引导算法的演化过程，可有效地提升知识引导的分布估计算法对于分布式异构混合流水车间调度问题的求解效率。

关键词: 分布估计算法, 混合流水车间调度, 分布式调度, 知识引导