在工業的批次製程中，最終的產品品質是由每個批次中的製程變數軌跡來決定的。因此，在建立預測品質模型時，我們有兩個重要的問題必須考慮。(1)每個批次過程中，製程變數軌跡的動態特性連續累積的來影響產品的最終品質；(2)每個製程變數在不同的時間階段下對產品的影響是不相同的，因此我們必須將這種累積影響與時間變化的特性做一個良好的解釋，以利我們建立準確的品質預測模型。
在傳統的多變數統計迴歸方法並無法良好將製程中變數的動態特性考慮進去，所以無法良好看出每個時刻對最終品質的重要性，因此在本篇論文中提出multiway group lasso (MGL)和multiway elastic net (MEN)兩種方法，均基於正規化(regularization)的變數選擇方法上，具有變數選擇及線性迴歸的特性，可以自動的調整出良好的品質預測模型迴歸係數，並將批次製程的上述兩種特性考慮進去。在透過適當的數據預處理方法後MGL和MEN相比於傳統的多變數統計迴歸方法均有更好的預測結果，並能對製程的物理特性做出良好的解釋，其中MEN相比於其他方法更顯現出準確的預測準確度。而在線上產品品質時所需的未來數據，本文推薦採用基於k-nearest neighhor (kNN)方法來估計未來數據。最終藉由本論文提出的數據預處理方法結合kNN估計方法，將MGL和MEN方法應用於射出成形機製程中，可以從結果看出MGL和MEN不僅提高了線上品質預測能力並且也提升了對製程機理過程的了解。

[1] P. Nomikos, J. MacGregor, Multi-way partial least squares in monitoring batch processes, Chemometrics and Intelligent Laboratory Systems, 30, (1995), 97–108.
[2] L. Wangen, B. Kowalski, A multiblock partial least squares algorithm for investi- gating complex chemical systems, Journal of Chemometrics , 3, (1989), 3–20.
[3] J. Westerhuis, T. Kourti, J. MacGregor, Analysis of multiblock and hierarchical PCA and PLS models, Journal of Chemometrics, 12, (1998), 301–321.
[4] S. Wold, N. Kettaneh, K. Tjessem, Hierarchical multiblock PLS and PC models for easier model interpretation and as an alternative to variable selection, Journal of Chemometrics, 10, (1996), 463–482.
[5] Y.H. Chu, Y.H. Lee, C. Han, Improved quality estimation and knowledge extraction in a batch process by bootstrapping-based generalized variable selection, Industrial and Engineering Chemistry Research, 43, (2004), 2680–2690.
[6] N. Lu, F. Gao, Stage-based process analysis and quality prediction for batch processes, Industrial and Engineering Chemistry Research, 44, (2005), 3547–3555.
[7] L. Xu, J.H. Jiang, H.L. Wu, G.L. Shen, R.Q. Yu, Variable-weighted PLS, Chemometrics and Intelligent Laboratory Systems, 85, (2007), 140–143.
[8] X. Wang, T. Park, K. Carriere, Variable selection via combined penalization for high-dimensional data analysis, Computational Statistics and Data Analysis, 54, (2010), 2230–2243.
[9] R. Tibshirani, Regression shrinkage and selection via the lasso, Journal of the Royal Statistical Society, Series B (Methodological), (1996), 267–288.
[10] M. Yuan,Y. Lin, Model selection and estimation in regression with grouped variables, Journal of the Royal Statistical Society. Series B (Methodological), (2006), 49–67.
[11] W.J. Fu, Penalized regressions: The bridge versus the lasso, Journal of Computational and Graphical Statistics, 7, (1998), 397-416.
[12] H. Zou, T. Hastie, Regularization and variable selection via the elastic net, Journal of the Royal Statistical Society: Series B (Statistical Methodology), 67, (2005), 301–320.
[13] J. Camacho, J. Pico, A. Ferrer, Bilinear modelling of batch processes. Part I: theoretical discussion, Journal of Chemometrics, 22, (2008), 299–308.
[14] Y. Yao, F. Gao, A survey on multistage/multiphase statistical modeling methods for batch processes, Annual Reviews in Control, 33, (2009), 172–183.
[15] P. Nomikos, J. MacGregor, Monitoring batch processes using multiway principal component analysis, AICHE Journal, 40, (1994), 1361–1375.
[16] L. Breiman, Stacked regressions, Machine Learning ,24 ,(1996), 49–64.
[17] B. Lennox, H. Hiden, G. Montague, G. Kornfeld, P. Goulding, Application of multi- variate statistical process control to batch operations, Computers and Chemical Engineering, 24, (2000), 291–296.
[18] H. Cho, K. Kim, A method for predicting future observations in the monitoring of a batch process, Journal of Quality Technology, 35, (2003), 59–69.
[19] I. Wasito, B. Mirkin, Nearest neighbour approach in the least-squares data impu- tation algorithms, Information Sciences, 169, (2005), 1–25.
[20] M. Stone, Cross-validatory choice and assessment of statistical predictions, Journal of the Royal Statistical Society. Series B (Methodological), (1974), 111–147.
[21] D. Montgomery, Design and analysis of experiments, 6th ed. Wiley, New York, 2005.
[22] C. Ündey, S. Ertunc, A. Cinar, Online batch/fed-batch process performance monitoring, quality prediction, and variable-contribution analysis for diagnosis, Industrial and Engineering Chemistry Research, 42, (2003), 4645–4658.