半导体生产线动态维护策略研究

发布时间：2018-08-04 21:26

【摘要】：半导体生产线结构复杂,具有典型的重入特点,加工产品种类多样,设备集成度高、造价昂贵。近年来半导体制造行业发展迅速,竞争激烈。合理的维护策略能够最大化设备价值,带来较高收益,提升企业市场竞争力。本文建立了设备生产和维护系统的马尔科夫决策过程MDP(markov decision process)模型。考虑可变的维护行为选择和随机的状态转移情况,研究半导体生产线设备的动态维护策略,获得使效益最大化的维护时刻和维护行为的综合。论文将此动态维护策略应用于一个典型的半导体制造过程-Mini-Fab模型,进一步研究模型中各单元的调度规则。对生产加工与维护抉择的耦合问题进行优化,得到综合最优的动态维护策略。首先,本文介绍了研究背景及意义,从工业界和学术界两方面分析国内外研究现状,详述研究内容和理论框架。其次,阐述了关于可靠性分析和维护研究的相关理论,以及生产调度的种类与判断指标,为后续研究提供理论支持。然后,在考虑可变的维护行为选择和随机的状态转移概率的情况下,建立了基于MDP模型的半导体设备动态维护策略的数学模型。以MDP模型模拟设备维护过程中状态转移情况,以设备效益作为目标函数,获得关于维护时刻和行为的综合方案。基于所建立数学模型的特点,引入带遗传交叉因子的粒子群算法求解,最终获得最优化动态维护策略。最后,将动态维护策略应用于Mini-Fab模型,并且进一步考虑系统生产加工过程,研究维护与加工冲突时的联合抉择问题和系统各单元的最佳调度规则,从而获得生产调度和动态维护方案的综合策略。
[Abstract]:Semiconductor production line has complex structure, typical reentry characteristics, various kinds of processed products, high integration of equipment and high cost. In recent years, semiconductor manufacturing industry has developed rapidly and the competition is fierce. Reasonable maintenance strategy can maximize the value of equipment, bring higher profits and enhance the market competitiveness of enterprises. In this paper, the Markov decision process MDP (markov decision process) model of equipment production and maintenance system is established. Considering the variable maintenance behavior selection and random state transition, the dynamic maintenance strategy of semiconductor production line equipment is studied, and the maintenance time and behavior synthesis to maximize the benefit are obtained. This dynamic maintenance strategy is applied to a typical semiconductor manufacturing process-Mini-Fab model to further study the scheduling rules of each unit in the model. The optimal dynamic maintenance strategy is obtained by optimizing the coupling problem between manufacturing and maintenance. First of all, this paper introduces the research background and significance, analyzes the domestic and foreign research status from industry and academia, and details the research content and theoretical framework. Secondly, the related theories of reliability analysis and maintenance research, as well as the types and judgment indexes of production scheduling are expounded, which provide theoretical support for the subsequent research. Then, considering variable maintenance behavior selection and random state transition probability, a mathematical model of dynamic maintenance strategy for semiconductor devices based on MDP model is established. The MDP model is used to simulate the state transition in the process of equipment maintenance, and the comprehensive scheme about the maintenance time and behavior is obtained by taking the benefit of the equipment as the objective function. Based on the characteristics of the established mathematical model, the particle swarm optimization algorithm with genetic crossover factor is introduced, and the optimal dynamic maintenance strategy is obtained. Finally, the dynamic maintenance strategy is applied to the Mini-Fab model, and further considering the production process of the system, the problem of joint selection in the conflict between maintenance and processing and the optimal scheduling rules for each unit of the system are studied. Thus, the integrated strategy of production scheduling and dynamic maintenance scheme is obtained.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN305;TP18

【参考文献】