GOALI: Intelligent Decomposition Heuristics for Scheduling Semiconductor Manufacturing Facilities

9613708尤兹索伊这一目标奖代表了工业和大学的联合研究努力，旨在为复杂的制造操作开发基于优化的近似调度程序。虽然基于优化的复杂制造系统调度方法并不新鲜，但它们在实践中的应用相对较少，主要是因为它们专注于计算复杂问题的准确解决方案。这项研究采取了不同的策略，使用启发式分解方法和机器学习的组合来为一类一般的作业车间开发有效的近似调度程序，这些作业车间包括顺序相关的安装时间、批处理和流水线。调查人员包括英特尔公司的一名高级研究员，他们将使用半导体晶片制造设施和组装测试设施作为他们研究的试验台。半导体制造的复杂性和行业面临的竞争压力使得有效的制造管理，特别是车间控制对公司的生存至关重要。车间控制的一个重要部分是调度通过工厂设备的物料移动，以获得最佳的系统性能。如果成功，这项工作有可能减少交货期的变异性，从而使公司能够保持较低的车间库存，对不断变化的市场状况做出快速反应，并向客户提供准确和有竞争力的交货期。工业伙伴在这项研究中的强大智力参与将进一步确保其与实际制造环境的相关性。

9613708 Uzsoy This GOALI award represents a combined industry-university research effort geared toward the development of optimization-based approximate scheduling procedures for complex manufacturing operations. While optimization-based scheduling approaches for complex manufacturing systems are not new, they have found relatively little use in practice, primarily because they have focused on exact solutions to computationally intractable problems. This research takes a different tack, employing a combination of heuristic decomposition methods and machine learning to develop effective approximate scheduling procedures for a general class of job-shops that encompass sequence dependent setup times, batching, and pipelining. The investigators include a senior researcher from Intel Corporation, and they will use semiconductor wafer fabrication facilities and assembly and test facilities as the testbeds for their research. The complexity of semiconductor manufacturing and the competitive pressures faced by the industry have made effective manufacturing management, in particular shop-floor control, essential to a company's survival. An important part of shop floor control is scheduling the movement of material through the equipment in the factory to obtain the best possible system performance. If successful, this work has the potential to reduce variability in lead times, thus allowing a company the ability to maintain lower floor inventories, respond rapidly to changing market conditions, and quote accurate and competitive lead times to their customers. The strong intellectual involvement of an industrial partner in this research will further ensure its relevance to practical manufacturing environments.