多种资源约束下集成电路芯片最终测试生产调度优化方法研究

发布时间：2018-04-02 22:05

本文选题：资源约束　切入点：批处理机　出处：《西南交通大学》2015年硕士论文

【摘要】：随着信息产业的蓬勃发展,半导体制造业已经成为国民经济的前驱发展产业。成本高、工序多、流程复杂为其特有属性。因此,如何提高设备利用率,对客户需求快速做出响应,并对有限的资源进行合理的分配利用成为了半导体制造业调度的目标。半导体集成电路生产线有前端工艺和后端工艺之分。前端工艺经过大量学者的研究,针对每个问题大都得到很好的优化方案,这就使得后端工艺流程中的问题凸显出来。对后端测试阶段集成电路芯片的优化调度可以有效地提高整个工艺流程的生产效益。后端测试阶段的调度问题大多数为NP-hard.再加上设备产能约束,以及工件加工过程中所需要的机器手、托盘等附属资源的约束,这就使得问题进一步复杂化。本文对多种资源约束下的半导体最终测试阶段的批处理机问题进行优化。我们首先针对单批处理机问题进行分析研究,建立问题的数学模型,并设计变邻域搜索算法(variable neighborhood search algorithm, VNS)、基于随机键的遗传算法(random key genetic algorithm, RKGA)以及基于批的插入算法(batch of insertion algorithm, BIA),最后通过数据仿真实验,将其与先前文献中的oven first batch first fit (OFBFF)启发式算法和一般遗传算法进行对比并对各算法的优劣性进行分析比较。之后本文将单批处理机调度问题扩展为同型号平行(identical parallel)批处理机调度问题,针对该问题我们根据多批处理机芯片生产流程的不同提出两种解决方案,其中每种解决方案都把问题分解为四个阶段,并分别设计出基于先组批再分配的批插入算法(first batching then machine- batch of insertion algorithm, FBTM-BIA)和基于先分配后组批的批插入算法(first machine then batching-batch of insertion algorithm, FMTB-BIA),通过仿真实验分析两种算法表现。
[Abstract]:With the rapid development of information industry, semiconductor manufacturing industry has become the leading industry of national economy.High cost, many processes, complex process is its unique attribute.Therefore, how to improve the utilization rate of equipment, to respond to customer demand quickly, and to allocate and utilize the limited resources rationally has become the target of semiconductor manufacturing scheduling.Semiconductor integrated circuit production line has front-end process and back-end process.After a lot of scholars' research, the front-end process gets a good optimization scheme for each problem, which makes the problems in the back-end process prominent.The optimal scheduling of IC chips in the back-end test stage can effectively improve the production efficiency of the whole process.Most scheduling problems in the back-end test phase are NP-hard.In addition, the constraints of equipment capacity, and the constraints of machine hands, pallet and other ancillary resources needed in the process of workpiece processing further complicate the problem.In this paper, the batch problem in the final testing phase of semiconductors with multiple resource constraints is optimized.First of all, we analyze and study the problem of single batch processor, and set up the mathematical model of the problem.Then this paper extends the single batch processor scheduling problem to the same type parallel parallel parallel batch processor scheduling problem. In view of this problem, we propose two solutions according to the different production processes of multi-batch processing machine chips.Each of these solutions breaks down the problem into four phases,The first batching then machine- batch of insertion algorithm (FBTM-BIA) and the first machine then batching-batch of insertion algorithm (FMTB-BIAA) are designed, respectively. The performance of the two algorithms is analyzed by simulation experiments.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN407

【参考文献】