半导体封装测试生产线调度方法研究
文献类型：学位论文
| 作者 | 姚丽丽 |
| 学位类别 | 博士 |
| 答辩日期 | 2014-05-19 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 导师 | 史海波 ; 刘昶 |
| 关键词 | 半导体封装测试 生产调度 投料控制 静态调度 重调度 |
| 其他题名 | Research on Scheduling in Semiconductor Assembly and Test Manufacturing |
| 学位专业 | 机械电子工程 |
| 中文摘要 | 本文以某半导体封装测试企业为研究背景，从投料控制和作业指派两个层次对半导体封装测试的生产过程进行优化研究，构建了完整的半导体封装测试生产调度框架体系；同时，针对各个层次、各个阶段的不同优化问题，从实际工程应用的角度出发，提出不同的优化解决算法，通过仿真测试，验证了各个新的算法的可行性和有效性。主要研究内容包括：封装测试制造系统调度分析、粗日投料控制研究、细日投料控制研究、静态调度研究和重调度研究等。具体的研究内容及创新成果概括如下：（1） 针对半导体封装测试生产特点，对其生产过程进行分析，从加工作业、加工设备、加工工艺、加工时间等四个方面对封装测试的的调度对象、调度参数、调度约束等生产调度要素进行阐述；同时，挖掘生产相关性能指标，并对各个指标进行量化描述，归类为成本型、效益型、固定型三种类型指标，并建立了封装测试特有的指标体系。封装测试制造系统分析为生产调度方法的研究提供了基本输入和研究目标。（2）首次将聚类的思想应用在投料控制中。针对半导体封装测试多品种投产的粗日投料控制问题，提出一种基于K-modes聚类的综合投料控制策略。该方法的主要思想是：以降低“改机”代价为目标，在投产之前对各个品种依据其不同改机属性进行聚类，然后在聚类的基础上采用基于品种平均和投产量平均的综合投料策略，进而确定每日的投产品种投产量。将本文所提出的算法与其他算法进行了比较，比较结果显示基于聚类的投料控制策略能够降低改机代价，提高设备利用率。（3）针对封装测试的细日投料控制问题，以粗日投料控制结果为输入，提出一种多属性决策的方法对类别内品种分别进行排序，主要考虑的因素有拖期率、交货期、客户重要度、订单接收日期等；针对具体投料时刻的确定问题，基于约束理论（Theory of Constrain， TOC），提出一种新的基于产能需求的细日投料控制策略，该策略能够在保证产能需求的基础上降低在制品（Work in Process, WIP）库存。（4）针对半导体封装测试的静态调度问题，提出一种新的产能限定混合流水车间模型（Capacity-Limit Flexible Flow-shop, CLFFS），并采用分解策略，分阶段对不同的优化问题进行解决。针对存有“改机”问题的阶段，提出一种新的限定分配优化调度方法，降低了生产过程中的“改机”代价，提高了设备利用率；针对“批准备单处理”生产阶段，提出一种新的预测开机控制优化调度方法，降低了Makespan目标，缩短了生产周期；针对“批准备-批处理”阶段，提出一种新的批量跨度两层控制调度方法，在保证Makespan目标的同时，降低了平均完工时间，从而降低了平均在制品值和平均等待时间。（5）针对封装测试的重调度问题，在对封装测试中引发重调度的因素、重调度策略、重调度方法和重调度性能评价等进行分析研究的基础上，建立了封装测试的重调度体系。同时，重点对封装测试中的重调度方法进行了研究。针对设备故障引起的重调度问题，提出一种新的右移结合更改路径的重调度方法，与传统方法相比，其具有更好的有效性和稳定性。针对Lot保留所引起的重调度，提出一种新的左移重调度方法，该方法能够缩短生产周期，使得重调度方案更为满足实际生产需求，提高生产效益。 |
| 索取号 | F253/Y35/2014 |
| 英文摘要 | Taking one ATM enterprise as study object, this paper researches on production scheduling from two aspects of lot release and job assignment, and builds a complete frame of ATM production scheduling. Many different optimization algorithms are proposed to solve different scheduling problems, and these algorithms have good feasibility and effectiveness in the application domain. The main content of the thesis includes: scheduling analysis in ATM, daily rough lot release control strategy, daily fine lot release control strategy, static scheduling methods, rescheduling methods. The specific content and innovative research results can be summarized as follows:(1) Aiming at the characteristics of ATM, this paper analyzes and summaries the production process of ATM, and discusses various scheduling factors such as scheduling object, scheduling parameters and scheduling constrains at four aspects, which are job, machine, process, and worktime. Meanwhile, a performance evaluation system for production scheduling in ATM is built, which consists of cost performance indices, benefit performance indices and fixed performance indices. Scheduling analysis in ATM provides the basic input and research goal for scheduling methods.(2) It is the first time that clustering algorithm is used to solve lot release problem. In order to address the daily rough lot release control problem in ATM system, a novel release control strategy based on product cluster analysis is proposed to reduce conversion cost in ATM. Firstly, a new quantity limited and feature weighted K-modes clustering algorithm is proposed to cluster the products in weekly scheduling. In the new algorithm, the number of bottleneck capacity types is taken as cluster number, and the number of points in each cluster class is limited. Cluster analysis is done according to the attributes which are relevant to conversion cost. Secondly, depending on the results of cluster analysis, daily release products and product-quantity is determined by the integrated strategy which has both considered product average distribution strategy and product-quantity average distribution strategy. Finally, by applying proven, the results show that the proposed strategy has the good feasibility and obvious advantage.(3) Daily fine release control strategy is studied based on the results of daily rough release control. Firstly, all the products in the same classification are sorted using multi-attribute prioritization algorithm, and the multi-attribute main includes: tardiness rate, order time, customer importance, received date and so on. Secondly, based on the theory of constrain (TOC), a new daily fine release control strategy is proposed to determine the release time, and the strategy can reduce work in process (WIP) inventory on the basis of ensuring capacity requirement.(4) In order to solve the static scheduling problems in ATM, a new model which is named capacity-limit flexible flow-shop (CLFFS) is proposed, and the paper adopts decomposition strategy (DS) to research on each scheduling algorithms in stages. Firstly, a novel scheduling algorithm is proposed for solving the problem of product-conversion, and the algorithm can reduce the cost of product-conversion and improve equipment utilization. Secondly, a novel scheduling algorithm is proposed for solving the problem of batch-preparation single-handling (BPSH), and the algorithm can minimize makespan and shorten production cycle time. Thirdly, a novel scheduling algorithm is proposed for solving the problem of batch-preparation batch-handling (BPBH), and the algorithm can reduce the average completion time.(5) For the rescheduling problems in ATM, the rescheduling framework for ATM is constructed based on analysis of rescheduling factors, rescheduling strategy, rescheduling method and rescheduling performance evaluation. This paper places great emphasis on the study of rescheduling method. Firstly, a new rescheduling method is proposed to solve the rescheduling problem of machine breakdown, and the algorithm has the better effectiveness and stability compared with the traditional algorithms. Secondly, a new rescheduling method is proposed to solve the rescheduling problem of lot reservation, and the algorithm can shorten production cycle time and improve production effectiveness. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 页码 | 131页 |
| 分类号 | F253 |
| 源URL | [http://ir.sia.ac.cn/handle/173321/14785]  |
| 专题 | 沈阳自动化研究所_数字工厂研究室 |
| 推荐引用方式 GB/T 7714 | 姚丽丽. 半导体封装测试生产线调度方法研究[D]. 中国科学院沈阳自动化研究所. 2014. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。