面向半导体制造的生产组织与设备维护安排联合优化问题研究

Both production and maintenance arrangement are very important activities in semiconductor manufacturing area, due to the complexity of the semiconductor manufacturing process and the high cost of the equipments utilized. Existing methods are that the production and the maintenance arrangements are separately made by different departments of an enterprise without considering the fact that there actually are significant interactions between them. The arrangements can not to be global optimization by the methods, and even lead to the conflicts between them and infeasibility of arrangements in practical execution. And also equipments are often maintained either overly or insufficiently, while production tasks may not be completed on time in accordance with the predetermined plan. Nowadays the above mentioned problems has severely weaken the competitiveness of many semiconductor manufacturing enterprises. Based on the analysis hereinbefore, and the fact that planning and scheduling are most critical to both production and maintenance arrangement, this propose aims to study the joint optimization of production planning and maintenance planning of semiconductor manufacturing, and the joint optimization of production scheduling and maintenance scheduling, while considering the characteristics of the semiconductor manufacturing process. We intend to build models for these two joint optimization problems based on deeply research into the relationship and interaction between production and maintenance arrangement. Additionally ,we will seek to develop effective and efficient solution techniques for both of these two problems. In summary, our project will provide a series of theories and techniques with respect to the joint optimization of production organization and maintenance arrangement while considering the particular complexity of the semiconductor manufacturing process, so as to provide guidance and means for the semiconductor manufacturing enterprises to solve the problems mentioned above.

半导体制造过程的复杂性和设备的昂贵使得其生产组织和设备维护安排成为非常重要的活动，现有的企业大都采用由生产部门和设备部门各自安排的方法，并未考虑生产组织和设备维护安排之间具有相互作用相互影响的关系，导致无论是生产组织还是设备维护安排都缺乏整体优化，甚至在实际操作中往往发生冲突而缺乏可行性，或设备过度维护或维护不足，或生产任务不能按计划完成，这些问题已成为目前严重影响企业竞争力的关键之一。本项目选择生产组织和设备维护安排的主要环节：计划与调度，针对半导体制造具有的典型特点，研究面向半导体制造的生产计划与设备维护计划的联合优化问题，以及生产作业与设备维护作业联合调度优化问题。通过研究半导体制造中生产组织和设备维护安排的相互作用和相互影响关系，找到影响的要素，对联合优化问题建模并设计出相应的求解方法。提出一套面向半导体制造的生产组织和设备维护安排联合优化的理论和方法为企业解决该类问题提供指导。

本项目针对半导体制造系统特点，研究面向半导体制造的生产计划与设备预防性维护计划的联合优化问题，以及生产作业与设备维护作业联合调度优化问题。对于生产计划与设备预防性维护计划的联合优化问题，以无缓冲和有缓冲的多机串行系统为对象，构建了考虑负荷影响与生产提前期的非线性关系的非固定提前期的生产计划与设备维护计划联合优化模型，设计了遗传算法以及基于时间分解的模拟退火算法进行求解；考虑半导体晶圆制造的重入多阶段特点，构建了设备负荷、预防性维护与生产提前期的多维非固定提前期产出函数，以此为基础提出了生产计划与预防性维护计划的联合优化模型，通过求解分析验证了该模型比已有模型更加符合实际，生成的计划具有更小的总成本。对于生产作业与设备维护作业联合调度优化问题，建立了柔性加工车间的以最优的维护周期和单位时间维护成本为目标的联合优化模型，提出了一种双重编码的遗传算法的求解方法；建立了最小化最大完工时间为优化目标可重入作业车间的联合优化数学模型，提出了混合粒子群算法的求解方法；建立了多机串行系统考虑工位可靠性与产品质量相互影响的维护调度模型，在此模型的基础上提出了基于滚动时域法的预防性机会维护策略；建立了多机串并行系统基于生产能力失效率指标的维护调度模型，提出了基于规则的预防性机会维护策略；建立了并行机联合优化模型，提出了基于蚁群算法的优化算法；建立了批处理机的联合优化模型，提出了启发式算法与改进的遗传算法相结合的求解方法；建立了批-离散机重入车间联合优化模型，设计了结合信息熵概念的自适应差分进化算法以及组合智能算法。以上研究成果揭示了这类复杂系统的联合优化中的基本问题和共性特征，提出了一套解决这类问题的有效方法和途径，为半导体企业解决该类问题提供了理论上和方法上的支持。