半导体制造业多目标中期生产计划优化模型研究

论文摘要

半导体制造业作为全球发展最快最先进的行业之一,其领导和充分利用最新的科技成果,是一个高风险与高利润并存的资本集中产业,其行业市场变化剧烈,机会稍纵即逝。半导体制造是一个流程高度复杂,资本高度密集的加工过程。随着半导体制造业的发展壮大,新的产业特点不断呈现,其产品种类繁多且生命周期不断缩短、多工厂、工序繁琐、物料清单复杂,设备的购买费用较高且交付周期较长,设备的柔性更高,对设备的利用率提出了较高要求。这些特点使得半导体制造业的生产计划尤为复杂。因此,企业必须缩短产品的市场反应时间,提高产品生产计划的制定水平,考虑各种相关因素,制定出有效的生产计划,来利用有限的生产资源,极大程度地满足客户需求,获得最大化利润。个优秀的企业生产计划应当充分考虑到种类繁多的产品特性,适应生产资源的灵活性、半导体制造业供应链的复杂性,还能对订单和生产能力的变化进行迅速反应。然而传统的制造业的生产计划工作大都是基于局部、地区性数据,通过电子表格和启发式算法来进行求解,生产计划的可行性和最优性很大程度上依赖于计划人员的经验和智慧,无法适应行业的新特点。企业迫切需要精确性更高、可行性更强的生产计划技术来处理大量数据,从而制定出更加科学合理的生产计划。本文紧扣半导体制造业的特点,根据半导体特有的生产流程,基于高级计划排程(Advanced Planning and Scheduling, APS)理论,全面考虑半导体封装测试工厂的物料、产能、资源利用率、优先级、库存、生产时间(Throughput Time, TPT)、产出率等多种约束限制,通过将半导体多阶段物料流问题建模为带转运的运输问题,完成了半导体封装测试工厂的中期生产计划多目标数学模型的构建。然后运用基于线性规划(Linear Programming, LP)的方法,借鉴了目标规划(Goal Programming, GP)的思想,采用ILOG CPLEX解算器对模型进行求解。最后根据某大型半导体封装测试工厂的实际数据进行实验。经过工厂实践证明,本文所构建的生产计划模型在投入使用后,大大提高了生产计划的效率和响应时间,减少了过多的库存水平和未满足的需求量,为解决实际的生产计划问题提供了非常有价值的解决方案。本文在最后进行了总结,提出了该研究的局限性,并对未来的研究方向提出了建议。

论文目录

• 摘要
• Abstract
• Chapter 1 Introduction
• 1.1 Research Background
• 1.2 Problem Raising
• 1.3 Significance of the Research
• 1.4 Research Methods and Thesis Structure
• Chapter 2 Relevant Theory and Literature Review
• 2.1 Overview of Production Planning
• 2.1.1 Introduction of Production Planning
• 2.1.2 Process of Production Planning
• 2.1.3 Hierarchies of Production Planning
• 2.1.4 Researches on Production Planning
• 2.1.5 Researches on Production Planning in Semiconductor Manufacturing Industry
• 2.2 Overview of APS Theory
• 2.2.1 Introduction of APS
• 2.2.2 Relations and Differences between APS and ERP
• 2.2.3 Researches on APS
• 2.3 Brief Summary
• Chapter 3 Problem Description
• 3.1 Production Process of Semiconductor Manufacturing
• 3.2 Materials Allocation in Semiconductor Manufacturing
• 3.3 Material Flows from FABs to ATMs
• 3.4 Material Flows in ATMs
• 3.5 Production Planning in Semiconductor Manufacturing
• 3.6 Main Task of Mid-term Production Planning in Semiconductor Manufacturing
• 3.7 Brief Summary
• Chapter 4 Mathematical Model of Mid-term Production Planning in Semiconductor Manufacturing
• 4.1 Advanced Planning
• 4.2 Multiple Objectives of Mid-term Production Planning
• 4.3 Modeling of Semiconductor Assembly and Test Procedures
• 4.4 Description of Mathematical Model
• 4.4.1 Model Assumption
• 4.4.2 Indices Description
• 4.4.3 Input Scalar Parameters Description
• 4.4.4 Modeling the Sort Schedule Routing Problem
• 4.4.5 Modeling the TRDI Schedule Routing Problem
• 4.4.6 Modeling the TRDI and Capacity Allocation Problem
• 4.4.7 Modeling the Down Binning, Cross Shipping Problem and Inventory
• 4.4.8 Weight Coefficient Description
• 4.5 Brief Summary
• Chapter 5 Implementation of the Mid-term Production Planning Model
• 5.1 Implementation Overview
• 5.1.1 System Objective
• 5.1.2 Key Inputs and Outputs
• 5.2 Running Process
• 5.3 Experiment Instance
• 5.4 Experiment Result
• 5.5 Implementation Effect
• 5.6 Brief Summary
• Conclusion
• Acknowledgement
• Publications during Graduate Years
• Reference
• Appendix A
• A.1 OPL code of getting input parameters data. Take the FAB data/index as examples
• A.2 OPL code of decision variables definition
• A.3 OPL code of objectives
• A.4 OPL code of constraints, take FAB constraints as examples