基于EDA-GA的置换流水车间调度算法

A Hybrid EDA with GA for the Permutation Flow Shop Scheduling Problem

摘要: 置换流水车间调度问题是工业工程中经典的组合优化问题，一般采用智能优化算法来求解该问题。作为一种新颖的优化算法，分布估计算法主要使用统计学习的方法指导搜索过程。然而，这种算法容易陷入到局部最优而出现过早收敛的现象。本文将分布估计算法与遗传算法结合，通过模糊逻辑控制来调节两种算法生成个体的比例。该算法有利于保持种群的多样性，避免了过早收敛。以Car类和Rec类算例进行测试，实验结果证实了本文所提出的混合算法的有效性。

Abstract: The permutation flow shop scheduling problem is a classical combinatorial optimization in indus-trial engineering. Population-based evolutionary algorithms (EA) are the common methods to solve this problem. As a novel EA, estimation of distribution algorithm (EDA) directs the algorithm search towards good solutions by statistical learning. However, this algorithm may trap into the local optimal and lead to the premature convergence. To overcome the drawback of EDA, this paper incorporates EDA with GA to address the PFSP. The participation rates of EDA and GA are adaptively regulated by fuzzy logic controller. The experiment results on the benchmarks validate the efficiency of the proposed algorithm.