摘 要

現行半導體工廠進行產品生產時若無透過先進製程控制系統進行製程控制，主要會依據生產機台績效來控制產品品質；當機台績效不穩定則會直接影響到產品品質，所以如何將不同規格產品於生產前分配到最適合生產的機台進行生產是一件十分重要的事。目前，若要控制產品品質在控制機台配貨方式往往需要花費大量人力，首先要先判斷前製程之產品品質並分析現行生產機台績效後才能進行手動派工讓貨進行生產，透過此手動作業才能讓產品品質維持在一定變異水準之內。

本研究希望運用統計預測模型將機台績效歷史資料建立預測模型來即時預測最新機台績效數據，透過每一筆新進點資料即可馬上預測出最新機台績效，以讓下一批貨進行生產機台選擇之參考，透過統計預測之數據，工程人員將不再需透過個人經驗來計算目前機台績效值並進行手動派貨作業；在本研究中共使用了五種統計預測模式，包含長期平均、移動平均(Moving Average)、加權指數移動平均(Exponentially Weighted Moving Average)、線性迴歸(Linear Regression)和多項式迴歸(Multi-Regression)模式，並透過誤差評估指標發現整體而言以移動平均(Moving Average)之預測方式最為準確，此預測數值產出後可直接提供給生產派貨相關系統進行參考運用；未來，並可動態的將不同產品品質派至最適機台進行生產，進而降低產品品質變異和縮短產品生產週期。

Abstract

Current semiconductor factory is based on process tool performance to control product quality. If there is no advanced process control system to control process performance, un-stable tool performance will impact product quality directly. It’s important that how to dispatch different specification product to suitable process machine. In current factory, we spend much manpower to judge product dispatch model. First of all, engineer has to check pre-process data to analyze process machine performance and then manually dispatch product to suitable machine. This methodology keeps product quality within standard variation by manual operation.

This research aims to build a tool performance predict model. This model is selected through evaluating several popular statistics forecasting models. We hope we can predict tool performance by new data input via this model. By the forecast form system, engineer does not need to dispatch product by experience, but by calculating model. In this research, we set up the model by 5 statistics forecasting models, including long-term average、Moving Average、Exponentially Weighted Moving Average、Linear Regression and Multi-Regression statistics models. Frome the results, we found Moving Average is the most suitable by error evaluation index analysis. In the future, we may use this methodology to improve product quality variation and shorten manufacturing cycle time.

參考文獻
中文部份
1. 侯凱中.(民92). 以限制理論為基礎的晶圓製造系統派工法則之研究, 機械工業雜誌 , 第15期，93-101.
2. 張耀仁.(民93). Run to Run 控制法則分析與比較. 機械工業雜誌 , 第246期，156-165.
3. 徐明照.(民93). Run to Run control簡介與APC framework架構.機械工業雜誌,第251期，35-45.
4. 高季安.(民89). EPC 在 IC 製程上的應用. 碩士論文,中原大學工業工程研究所.
5. 游政雄.(民89). 化學機械研磨之類神經網路式Run-to-Run製程控制. 碩士論文,國立中興大學機械工程學系.
6. 林家瑞.(民93). 化學機械研磨移除厚度之批片雙模式自適應控制. 碩士論文,國立交通大學機械工程所.
7. 王宗一.(民94). 多代理人系統在建構 APC 作業平台之研究. 碩士論文, 國立成功大學工程科學系專班.
8. 王俊凌.(民94). 半導體先進批次製程控制系統平台設計與實作. 碩士論文,中央大學資訊管理學系碩士在職專班.
9. 陳順宇.(民89). 基礎統計量. 華泰書局.
10. 王勝弘.(民92). 派工法則於縮短DRAM 廠生產週期時間之探討—以A公司導入自動化派工系統為例. 碩士論文, 中央大學管理學院高階主管企管碩士班.
11. 陳陸國.(民82). 晶圓製造廠各機種派工法則之系統化設定. 碩士論文,交通大學工業工程研究所.
12. 吳忠武(民93). 現代統計學二版 . 華泰文化事業股份有限公司.
13. 林傑斌(民91). 資料採掘與OLAP理論與實務. 文魁出版社.
14. 顧力栩(民97) 人工智慧:智慧型系統導論(第二版). 全華圖書股份有限公司.
15. 曹永誠(民94) 先進製程控制技術(APC)導論. 電機月刊雜誌,第174期，1-14.
英文部份
Arnon, M. H., Enrique, D. C.,& James, M. (2000). Run to Run Control in Semiconductor Manufactuing, Lewis Publishers, Taylor & Francis.

Boning, D., Moyne, W.,& Smith, T. (1996). Run by run control of Chemical-Mechanical Polishing. Institute of Electrical and Electronics Engineers Transactions on Semiconductor Manufacturing, 19(4), 307-314.

Borror, C. M.,Nong, Y.,& Qiang, C. (2004). EWMA forecast of normal system activity for computer intrusion detection. Reliability, Institute of Electrical and Electronics Engineers Transactions on Semiconductor Manufacturing, 4(53), 557-566.

Castillo, E.,& Hurwitz, A. (1997). Run to run process control: Literature review and extensions. Quality Technology, 29(2), 184-196.

Chang, Z., Hao, D.,& Baras, J. S. (2003). Performance Evaluation of Run to Run Control Methods in Semiconductor Process .Control Applications, 2, 841-848.

Chen, S. F.,& Yin, Y. (2007). Research on the prediction model to the highway transportation demand based on moving average and grey theory. Grey Systems and Intelligent Services, IEEE International Conference, Nanjing.

Dong,J. R. (2007). Combining Stock Market Volatility Forecasts Using an EWMA Technique. Wireless Communications, Networking and Mobile Computing, The 3rd International Conference on Wireless Communications, Networking and Mobile Computing (WiCOM 2007) , Shanghai ,China.

Fukuoka, H., Konishi, S., & Yasanaga, M. (2000). Forecasting models for road-vehicle communication traffic in ITS. Paper presented at the meeting of Intelligent Transportation Systems,2000, Dearborn, MI, USA.

Hu, A., Ingolfsson, A.,& Sachs, E. (1992). Run by run process control:Performance Benchmarks. Semiconductor Manufacturing Science Symposium, Symposium conducted at the meeting of the IEEE/SEMI Internationnal Semiconductor Manufacturing Scicence , San Francisco, CA, USA.

Hurwitz, A.,Moyne, J.,& Ning, Z. (1996). A comparative analysis of run to run control algorithms in the semiconductor manufacturing industry. Institute of Electrical and Electronics Engineers / Advanced Semiconductor Manufacturing Conference, Cambridge, MA, USA. 375-381.

Jingyi, L. (2006). Forecasting Volatility Using Competitive Time Series models-Evidence from Foreign Exchange Market. Service Systems and Service Management , 1, 652-657.

Le, A. T., & Mazengia, D. H. (2008). Forecasting spot electricity market prices using time series models. Sustainable Energy Technologies, Institute of Electrical and Electronics Engineers International Conference, Singapore, 1256-1261.

Li, H., Hung, R.,Shanmugasundram, A., Schwarm, A., Sarfaty, M., Seamons, M.J.,et al. (2002). Advance Process Control solutions for semiconductor manufacturing Sarfaty. Advanced Semiconductor Manufacturing 2002 Institute of Electrical and Electronics Engineers / Advanced Semiconductor Manufacturing Conference and Workshop, Santa Clara, California ,101-106.

Lin, B., Patel, N. S.,& Yuoh, G. (2003). Poly Silicon Deposition Process Improvement on 300 mm Wafers. Institute of Electrical and Electronics Engineers Semiconductor Manufacturing ,Dallas, USA, 119-112.

Manabu, F.,& Shin, L. (2005). Development of equipment –installed APC system and critical dimension control technology of gate-hard-mask etching using its system. Semiconductor Manufacturing. 139-142.

Mehaoua, A., Tabbance, N.,& Tabbane, S. (2004). Autoregressive, movingaverage and mixed autoregressive-moving average processes for forecasting QoS in ad hoc networks for real-time service. Vehicular Technology Conference, Tunisia, 2580-2584.

Patel, N. S. (2004). Lot allocation and Process Control in Semiconductor Manufacturing – A dynamic game approach. Decision and Control, 4, 14-17.

Xia, B. and Zhao, C. (2009). The Application of Multiple Regression Analysis Forecast in Economical Forecast : The Demand Forecast of our country industry lavation machinery in the year of 2008 and 2009. Knowledge Discovery and Data Mining. Second International Workshop, Moscow , 405-408.