臺灣博碩士論文加值系統

在晶圓加工過程中，晶柱(Ingot)經切片(Slicing)後需透過圓邊(Edge Grinding)製程來防止後績加工過程中產生碎邊(Chipping)或汙染微粒。然而，因連績且大量的生產過程，以現行的品檢及預防維護的方式，容易造成人為誤判，導致研磨砂輪更換時機的延誤，使得研磨機台容易因問題累積而加工出報廢品或產品需要重工(rework)，進而提高加工成本。本研究透過在加工機台上安裝加速規與參克風，透過訊號擷取與分析，探討矽晶圆圓邊製程中時域、頻域訊號，將時頻域訊號針對機台加工特性擷取重要特徵建立特徵資料集，利用機器學習回歸模型建模，並用晶圓品質最為標籤標記資料，藉預測晶片倒角值，間接監測砂輪磨耗狀況，提早發現機台問題，在下次加工前採取對應手段。

In the process of wafer processing, the ingot must be sliced and then cut through the edge grinding process to prevent chipping or contamination particles during the subsequent processing. However, due to the continuous and large-scale production process, the current quality inspection and preventive maintenance methods are easy to cause human misjudgment, resulting in delays in the replacement of the grinding wheel, and the grinding machine is likely to process scrap or products due to accumulation of problems. Rework is required, which in turn increases processing costs. In this research, through the installation of accelerometers and parameters on the processing machine, through signal acquisition and analysis, the relationship between time domain and frequency domain signals, grinding wheel wear and wafer quality in the silicon wafer rounding process is explored, and early discovery For machine problems, predict the quality of wafers and take corresponding measures before the next processing.

摘要 i
Abstract ii
目錄 iii
圖目錄 vi
表目錄 xi
第一章. 緒論 1
1.1前言 1
1.2文獻回顧 2
1.2.1刀具狀態監控(Tool Condition Monitoring, TCM) 2
1.2.2 感測器與訊號擷取 3
1.2.3訊號處理與特徵擷取 6
1.2.4 機器學習應用於刀具監測 11
1.2.5 超參數選擇與超參數最佳化方法 13
1.3研究動機與目的 14
1.4論文架構 16
第二章. 研究方法 17
2.1奈奎斯取樣定理(Nyquist Sampling Theorem) 17
2.2快速傅立葉(Fast Fourier Transform ,FFT) 18
2.3決策樹(Decision tree) 22
2.4隨機森林(Random forest, RF) 23
2.5極限隨機樹(Extremely randomed tree, ET) 25
第三章. 實驗規劃 26
3.1實驗設備 27
3.1.1實驗機台 27
3.1.2砂輪規格 28
3.1.3晶圓規格 29
3.2量測系統設計 30
3.2.1硬體架設 32
3.2.2軟體設計 33
3.2.3參數設定 34
3.2.4主軸對應加速規座標系定義與量測系統安裝 36
3.3實驗設計 38
第四章. 實驗結果與討論 42
4.1資料集處理與分析 44
4.2訊號處理與特徵擷取 51
4.2.1訊號處理 51
4.2.2特徵擷取 61
4.3機器學習模型建立 65
4.3.1 Data Set劃分 65
4.3.2隨機森林數特徵排序與篩選 67
4.3.3隨機森林樹預測模型建立 70
4.3.4極限隨機樹特徵排序與篩選 73
4.3.5極限隨機樹預測模型建立 76
4.4倒角值允收規格 79
4.4.1倒角值直方圖統計 80
4.4.2 預測區間與T分佈正規化 81
第五章. 結論 83
5.1機器學習模型比較 83
5.2外掛量測系統評估 84
5.3圓邊站品管能力分析 91
5.3.1圓邊站作業流程 91
5.3.2 現行品管能力分析 92
5.3.3品管能力綜合分析(倒角值) 94
5.4品質預測GUI界面 96
第六章. 未來規劃 98
參考文獻 99
附錄一 加速規規格表 102
附錄二 麥克風規格表 106
附錄三 紅外線近接開關規格表 108
附錄四 擷取卡規格表 109
NI-9234 DAQ 109
NI-9223 DAQ 111
附錄五 機箱規格表 112
NI-9178 CDAQ 112
附錄六 桌上型小電腦規格表 113
附錄七 T Distribution Probability Table[27] 114
附錄八 倒角值實際量測與模型預測誤差分析 116

[1] M. Bhuiyan, I. Choudhury. 13.22—Review of sensor applications in tool condition monitoring in machining. Comprehensive Materials Processing 13: 539-569,2014.
[2] Y.Lei, B.Yang, X.Jiang , F. Jia, N.Li & A. K.Nandi. Applications of machine learning to machine fault diagnosis: A review and roadmap. Mechanical Systems and Signal Processing, 138, 106587.L,2020.
[3] I. Inasaki & K. Okamura. Monitoring of dressing and grinding processes with acoustic emission signals. CIRP Annals,34(1), 277-280,1985.
[4] P. O. Junior, P. R. Aguiar, C. R. Foschini, T. V. França, D. M. Ribeiro, F. I. Ferreira, & E.C. Bianchi. Feature extraction using frequency spectrum and time domain analysis of vibration signals to monitoring advanced ceramic in grinding process. IET Science, Measurement & Technology,13(1), 1-8, 2019.
[5] K.Kannan, N. Arunachalam. Grinding wheel redress life estimation using force and surface texture analysis. Procedia CIRP 72: 1439-1444,2018.
[6] 黃詩涵，主軸振動與聲音訊號於晶圓磨削之應用研究，國立清華大學奈米工程與微系統所，碩士論文，2019。
[7] S.Cho , S.Binsaeid, & S. Asfour. Design of multisensor fusion-based tool condition monitoring system in end milling.The International Journal of Advanced Manufacturing Technology,46(5), 681-694, 2010.
[8] M.Hamadache, J. H.Jung, J.Park, & B. D.Youn. A comprehensive review of artificial intelligence-based approaches for rolling element bearing PHM: shallow and deep learning. JMST Advances,1(1), 125-151, 2019.
[9] X.Li and Y. Zhejun. Tool wear monitoring with wavelet packet transform—fuzzy clustering method. Wear 219(2): 145-154,1998.
[10] D.G.Manolakis, V.K. Ingle .Applied Digital Signal Processing ,Cambridge, New York, 2011.
[11] F.Al-Badour, M.Sunar, & L.Cheded. Vibration analysis of rotating machinery using time–frequency analysis and wavelet techniques.Mechanical Systems and Signal Processing,25(6), 2083-2101, 2011.
[12] Y.Yabin, & Z.Liqiang. Process Research Based on Wavelet Analysis in Chatter Monitoring. In 2019 3rd International Conference on Electronic Information Technology and Computer Engineering (EITCE)(pp. 708-711)IEEE, 2019.
[13] E.Kolman, & M. Margaliot. Are artificial neural networks white boxes?. IEEE Transactions on Neural Networks, 16(4), 844-852, 2005.
[14] L.Breiman. Random forests. Machine learning, 45(1), 5-32,2001.
[15] X.Xue, C.Li, S.Cao, J.Sun, & L.Liu. Fault diagnosis of rolling element bearings with a two-step scheme based on permutation entropy and random forests. Entropy, 21(1), 96,2019.
[16] S.V.Dahe, G. S.Manikandan, R. Jegadeeshwaran, , G.Sakthivel, & J.Lakshmipathi. Tool condition monitoring using Random forest and FURIA through statistical learning. Materials Today: Proceedings, 2021.
[17] B. S.Yang, X.Di, & T.Han. Random forests classifier for machine fault diagnosis. Journal of mechanical science and technology,22(9), 1716-1725, 2008.
[18] J.Bergstra, & Y.Bengio. Random search for hyper-parameter optimization. Journal of machine learning research, 13(2),2012.
[19] A.V.Oppenheim, A.S.Willsky, S. H.Nawab. Signals and systems, PRENTICE HALL, NEW JERSEY,1997.
[20] H.T.Lin.Machine Learning Techniques, Blending and Bagging,2018.
[21] T.Han, D.Jiang, Q.Zhao, L.Wang, &K.Yin. Comparison of random forest, artificial neural networks and support vector machine for intelligent diagnosis of rotating machinery.Transactions of the Institute of Measurement and Control,40(8), 2681-2693,2018.
[22] G.Xu, M.Liu, Z.Jiang, D.Söffker, & W. Shen. Bearing fault diagnosis method based on deep convolutional neural network and random forest ensemble learning.Sensors,19(5), 1088,2019.
[23] M.Hamadache, J. H.Jung, J.Park, & B. D.Youn. A comprehensive review of artificial intelligence-based approaches for rolling element bearing PHM: shallow and deep learning. JMST Advances,1(1), 125-151,2019.
[24] J.Chen, H.Chen, J.Xu, J.Wang, X.Zhang, & X.Chen. Acoustic signal-based tool condition monitoring in belt grinding of nickel-based superalloys using RF classifier and MLR algorithm.The International Journal of Advanced Manufacturing Technology,98(1), 859-872,2018.
[25] M.Cerrada, G.Zurita, D.Cabrera, R. V.Sánchez, M.Artés, & C.Li. Fault diagnosis in spur gears based on genetic algorithm and random forest. Mechanical Systems and Signal Processing, 70, 87-103,2016.
[26] M.Same, G.Gandubert, G.Gleeton, P.Ivanov, & R. Landry. Simplified welch algorithm for spectrum monitoring. Applied Sciences, 11(1), 86,2021.
[27] Walpole, Ronald E., et al. Probability and statistics for engineers and scientists. Vol. 5. New York: Macmillan, 1993.