摘要
针对汽车制造过程中的总装车间混流排产规划问题,为避免生产系统信息资源的浪费,在对总装车间混流排产规划问题分析后建立了切换调整费用最小化、装配线空闲/停线时间最小化、物料消耗均衡化的多目标优化模型,并据此提出了一种以遗传算法、差分算法为底层算法库,结合Q-learning强化学习上层策略的超启发式算法,同时根据问题特性设计了一种基于排列组合的染色体编码方式,最后通过该算法与标准GA(遗传算法)、DE(差分算法)对不同批次订单的求解效率、质量进行分析比较,证实了该算法的优越性。
To tackle the scheduling of mixed-flow automotive assembly lines,this paper focuses on solving the problem of the waste of information in the current research situation.After analyzing the characteristics of the mixed production scheduling problem in the automotive assembly line,a hyper-heuristic algorithm based on the Q-learning reinforcement learning strategy is proposed.Therefore,a multi-objective optimization model was established based on this algorithm,to minimize the switching adjustment cost and the idle/stop time of the assembly line,to balance the material consumption.According to the characteristics of the problem,a chromosome encoding method based on permutation and combination is proposed.Finally,the algorithm is compared with the standard GA(genetic algorithm)and DE(differential algorithm)in terms of the efficiency and quality of different orders,which confirms the algorithm’s effectiveness and superiority.
作者
卢梓扬
盛步云
王辉
李晓芳
LU Ziyang;SHEN Buyun;WANG Hui;LI Xiaofang(School of Mechanical and Electronic Engineering,Wuhan University of Technology,Wuhan 430070,China;Hubei Key Laboratory of Digital Manufacturing,Wuhan 430070,China;Hubei Institute of Aerospace Metrology and Testing Technology,Xiaogan 432000,China)
