臺灣博碩士論文加值系統

即時監控分析方法對於製程良率提升之研究－以製造業W公司為例
學生：張哲瑜 指導教授：賴國華 博士
元智大學資訊工程學系
摘要
製程的良率是工廠所關心的問題。因為良率的問題會直接影響到產品的品質與最終的獲利。當產品品質發生異常的情況時，工程師們需執行後續的修正措施及錯誤檢測，並根據相關的生產紀錄做異常分析與改善。這些方式都是後續的補救方式，直接影響到了生產效率。所以，若能利用每條產線上，過去生產過的產品參數，加上大數據分析技術，預先找出造成產品異常的可能因素並且提早進行調整，將有助於改善製程良率、降低生產成本及減少人力成本。
本論文將以過去生產過程中有缺陷的產品為例。第一步先運用分類演算法判斷新產品參數設定值跟過去生產過的產品參數設定值何者最相似。第二步就是跟過去正常產品參數進行比對，如果變異值太大，我們就判斷目前產品生產異常。第三步說明如何收集跟建立我們比對的Knowledge Database。本研究將著重於及時機台參數異常偵測，並且將分析比對的方法加入多核心平行化運算。如此一來就算處理大量資料的機台參數，也不用擔心分析速度的問題。最後就是如何將過去正常的產品參數，建立成我們比對的Knowledge Database，達到模型不斷更新，比對將會更有效率跟準確。

關鍵詞: 平行運算、分類演算法、錯誤偵測、良率改善

Using real-time system monitor method to enhance Manufacturing yield
-An Application to Manufacturing Company W
Student: Chang, Che-Yu Advisor: Dr. K.Robert Lai
Department of Computer Science and Engineering
Yuan Ze University
ABSTRACT
The process yields are issues that every factory is concerned with. Because the issue of the process yields will affect production quality and final profit directly. When production quality has defects, engineers should execute error detection and corrective measures, and analyze the abnormality and improve them based on the related production records. These are the following remedies which affect the productivity directly. Therefore, if we can use production parameters which are produced before in each production line and the technique of big data analysis to find the possible factors which cause product defects and then adjust them in advance, it will help us improve the process yields, and reduce production and labor costs.
This thesis will take the defective products which are produced before for illustration. Firstly, to use classification algorithms to decide which parameter settings of new products are the most familiar to those produced before. Secondly, compare with the normal production parameters which are produced before, if the variance of new products is too large, we can learn that the products which are produced presently are defective. Thirdly, to explain how we collect and establish the Knowledge Database. This research will emphasize the real-time production line parameter fault detection, and add the measures of analysis and comparison to multi-kernel parallel computing. In this manner, even if we deal with the production line parameter which has a great amount of data, we don’t have to worry about the analysis rate problem. Finally, we can make the comparison more efficient and accurate by how we use the normal parameters produced before to establish the Knowledge Database to reach the constant update of models.
Keyword: Parallel Computing, Classification, Fault Detection, Yield Improvement

目錄
摘要 iii
ABSTRACT iv
誌 謝 v
目錄 vi
表目錄 viii
圖目錄 ix
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機與重要性 1
1.3 論文架構 2
第二章 文獻回顧 3
2.1 擴散板製程 3
2.2 資料探勘方法: 4
2.3 Manufacturing Fault Detection 4
2.4 Distance Base Fault Detection 5
第三章 研究方法 8
3.1 定義問題 8
3.2 Knowledge Database建立 9
3.3 即時監控方法流程 10
3.3.1 預測相似料號模型 12
3.3.2 平行化處理參數比對(多核心平行運算版本) 13
3.3.3 如何決定每個料號的最佳門檻值"α" 13
第四章 研究實證 14
4.1 找尋KNN最佳距離公式 14
4.2 演算法評測 &真實資料Fault detection 15
4.2.1 波紋NG—同料號比對模式 15
4.2.2 缺料NG –跨料號比對模式 23
第五章 結論 30
第六章 未來展望 30
第七章 參考文獻 31

 

第七章 參考文獻

1. Berry, M.J. and G. Linoff, Data mining techniques: for marketing, sales, and customer support. 1997: John Wiley & Sons, Inc.
2. Cover, T.M. and P.E. Hart,Nearest neighbor pattern classification. IEEE Transactions on Information Theory, 1967. 13(1): p.21-27
3. Tan, S., Neighbor-weighted k-nearest neighbor for unbalanced text corpus. Expert Systems with Applications, 2005. 28(4): p. 667-671.
4. Ison, A. and C.J. Spanos. Robust fault detection and fault classification of semiconductor manufacturing equipment. Semiconductor Manufacturing. 1996. Citeseer in Proc. Int. Symp.
5. He, Q., A. Tuskegee, and J. Wang, Statistics Pattern Analysis and Its Application to Semiconductor Process Monitoring.
6. Kuan, Y.W., L.C. Chew, and L.W. Jau. Method for proposing sort screen thresholds based on modeling etest/sort-class in semiconductor manufacturing. in 2008 IEEE International Conference on Automation Science and Engineering. 2008. IEEE.
7. Lee, L.L., C.D. Schaper, and W.K. Ho, Real-time predictive control of photoresist film thickness uniformity. IEEE Transactions on Semiconductor Manufacturing, 2002. 15(1): p. 51-59.
8. Hess, C. and L.H. Weiland, Extraction of wafer-level defect density distributions to improve yield prediction. IEEE transactions on semiconductor manufacturing, 1999. 12(2): p. 175-183.
9. Ali, I. and Y.-T. Chen, Design quality and robustness with neural networks. IEEE Transactions on Neural Networks, 1999. 10(6): p. 1518-1527.
10. He, Q.P. and J. Wang, Fault detection using the k-nearest neighbor rule for semiconductor manufacturing processes. IEEE transactions on Semiconductor manufacturing , 2007. 20(4): p. 345-354.
11. Verdier, G. and A. Ferreira. Fault detection with an adaptive distance for the k-nearest neighbors rule. in Computers & Industrial Engineering, 2009. CIE 2009. International Conference on. 2009. IEEE.
12. He, Q.P. and J. Wang, Large-scale semiconductor process fault detection using a fast pattern recognition-based method. IEEE Transactions on Semiconductor Manufacturing, 2010. 23(2): p. 194-200.
13. Zhou, Z., C. Wen, and C. Yang, Fault Detection Using Random Projections and k-Nearest Neighbor Rule for Semiconductor Manufacturing Processes. Semiconductor Manufacturing, IEEE Transactions on, 2015. 28(1): p. 70-79.
14. Wu, X., et al., Top 10 algorithms in data mining. Knowledge and information systems, 2008. 14(1): p. 1-37.
15. Huang, Y.-T., et al. Automatic virtual metrology system design and implementation. in 2008 IEEE International Conference on Automation Science and Engineering. 2008.
16. Huang, Y.-T., F.-T. Cheng, and M.-H. Hung. Developing a product quality fault detection scheme. in Robotics and Automation, 2009. ICRA'09. IEEE International Conference on. 2009. IEEE.