線上缺陷檢測是爲面板品質把關的重要程序。在線上缺陷檢測中，又以陣列製程的缺陷檢測最為重要，因為陣列製程為面板三大製程之首。Test for Open/Short (TOS) 是陣列製程中線上缺陷檢測的方法之一。TOS主要是檢驗面板的電性缺陷。目前TOS機臺所遇到的瓶頸是當缺陷訊號不明顯時，TOS機臺沒有辦法正確檢出面板缺陷，因而有缺陷漏檢（型一誤差）的情況。型一誤差發生時不僅會降低面板良率，還會增加生產時間及成本。在目前的面板製程中，工程師會將面板重複進行兩次線上缺陷檢測，期望能夠透過第二次檢測提高缺陷被檢出之機會，減少型一誤差之發生。然而當缺陷在二次檢測（二次回測）被檢出時，往往都已經進行到後段製程，在這等待缺陷進入到二次檢測的過程中，多半都已浪費許多時間及生產成本。因此，為了解決目前TOS型一誤差過高之現象，本研究發展出一套兩階段的線上缺陷自動檢測演算法，透過分析面板在TOS機臺通電後之電壓值變化，期望能及早檢出缺陷並修復。第一階段利用支援向量機建構訊號分類模型，第二階段是通過FIR（Finite impulse response, FIR）濾波器進行缺陷位置判定。實證結果顯示，此演算法可以讓TOS機臺有效減少42.8%之型一誤差，提升陣列製程的良率，並取代陣列製程進行二次回測的必要，進而降低生產的時間及成本。

The inline defect detection in array process is the most important one because the array process is the first process. Test for Open/Short (TOS) is one of the inline defect detection methods in array process. TOS aims to detect electrical defects. The shortcoming of TOS is that it tends to judge the defective signal as normal while the strength of defect is weak, which is so-called type I error. The type I error will not only decrease the yield but also increase the time and cost of production. In TFT-LCD manufacturing process, the engineers now use the second-round defect detection to double check whether the panel exists defect or not. This kind of second-test will take a lot of time until the defect is detected, making the time cost extremely high. This paper has considered the problem of type I error in TOS. We propose a two-phase automatic defect detection algorithm. The algorithm directly works on the voltage signals captured by TOS. Phase 1 aims to do signal classification based on SVM, while Phase 2 aims to do defect detection for both strong and weak defective panel by FIR filter. Experimental results show that the proposed method can reliably reduce the rate of type I error by 42.8% in TOS, making the yield becomes better and saving the time and cost of production in array process.

Chapter 1 Introduction 1
1.1 Background and motivation 1
1.2 Purpose 2
1.3 Structure 3
Chapter 2 Literature Review 4
2.1 Inline defect detection in array process 4
2.1.1 Inline defect detection methods 4
2.1.2 Fundamentals of TOS defect detection 6
2.2 Application for filter in defect detection 8
2.3 Application for one-dimensional time-series signal analysis 10
Chapter 3 Methodology 12
3.1 Problem Definition 15
3.2 Phase 1 : Signal Classification 15
3.2.1 Data preparation 15
3.2.1.1 Data collection and examination 15
3.2.1.2 Data cleaning 16
3.2.1.3 Feature extraction 17
3.2.1.4 Data segmentation 17
3.2.2 Modeling 18
3.2.2.1 Model Building 18
3.2.2.2 Model Validation 20
3.2.3 Result assessment 20
3.3 Phase 2 : Defect Location Identification 21
3.3.1 Data preparation 21
3.3.2 Modeling 21
3.3.3 Result assessment 27
Chapter 4 Empirical study 29
4.1 Problem definition 30
4.2 Phase 1 : Signal Classification 30
4.2.1 Data preparation 30
4.2.1.1 Data collection and examination 31
4.2.1.2 Data cleaning 35
4.2.1.3 Feature extraction 39
4.2.1.4 Data segmentation 40
4.2.2 Modeling 40
4.2.2.1 Model building 40
4.2.2.2 Model validation 44
4.2.3 Result assessment 46
4.3 Phase 2 : Defect Location Identification 47
4.3.1 Data preparation 47
4.3.2 Modeling 47
4.3.3 Result assessment 52
4.4 Online application 54
Chapter 5 Conclusion and Future Work 57
5.1 Conclusion and contribution 57
5.2 Future research 57
Reference 59

Abeysundara, H., Hamori, H., Matsui, T., and Sakawa, M. (2013), "Defects detection on TFT lines of flat panels using a feed forward neural network," Artificial Intelligence Research, Vol. 2, No. 4,pp. 1.
Barhatte, A. S., Ghongade, R., and Tekale, S. V. (2016), "Noise analysis of ECG signal using fast ICA," Proceedings of 2016 Conference on Advances in Signal Processing (CASP).
Chawla, N. V., Bowyer, K. W., Hall, L. O., and Kegelmeyer, W. P. (2002), "SMOTE: synthetic minority over-sampling technique," Journal of Artificial Intelligence Research, Vol. 16,pp. 321-357.
Chen, L.-C. and Kuo, C.-C. (2007), "Automatic TFT-LCD mura defect inspection using discrete cosine transform-based background filtering and ‘just noticeable difference’quantification strategies," Measurement Science and Technology, Vol. 19, No. 1,pp. 015507.
Cheng, F.-T., Hsieh, Y.-S., Zheng, J.-W., Chen, S.-M., Xiao, R.-X., and Lin, C.-Y. (2017), "A scheme of high-dimensional key-variable search algorithms for yield improvement," IEEE Robotics and Automation Letters, Vol. 2, No. 1,pp. 179-186.
Cortes, C. and Vapnik, V. (1995), "Support-vector networks," Machine Learning, Vol. 20, No. 3,pp. 273-297.
Hamori, H., Sakawa, M., Katagiri, H., and Matsui, T. (2010), "Fast non-contact flat panel inspection through a dual channel measurement system," Proceedings of 40th International Conference on Computers and Industrial Engineering (CIE).
He, Q., Yan, R., Kong, F., and Du, R. (2009), "Machine condition monitoring using principal component representations," Mechanical Systems and Signal Processing, Vol. 23, No. 2,pp. 446-466.
Hsu, C.-W. and Lin, C.-J. (2002), "A comparison of methods for multiclass support vector machines," IEEE Transactions on Neural Networks, Vol. 13, No. 2,pp. 415-425.
Jazi, A. Y., Liu, J. J., and Lee, H. (2012), "Automatic inspection of TFT-LCD glass substrates using optimized support vector machines," IFAC Proceedings Volumes, Vol. 45, No. 15,pp. 325-330.
Knerr, S., Personnaz, L., and Dreyfus, G. (1990), "Single-layer learning revisited: a stepwise procedure for building and training a neural network," Neurocomputing: algorithms, architectures and applications, Vol. 68, No. 41-50,pp. 71.
Kumar, K. S., Yazdanpanah, B., and Kumar, P. R. (2015), "Removal of noise from electrocardiogram using digital FIR and IIR filters with various methods," Proceedings of 2015 International Conference on Communications and Signal Processing (ICCSP).
Li, B., Goddu, G., and Chow, M.-Y. (1998), "Detection of common motor bearing faults using frequency-domain vibration signals and a neural network based approach," Proceedings of 1998 American Control Conference.
Liu, Y.-H., Wang, C.-K., Ting, Y., Lin, W.-Z., Kang, Z.-H., Chen, C.-S., and Hwang, J.-S. (2009), "In-TFT-array-process micro defect inspection using nonlinear principal component analysis," International Journal of Molecular Sciences, Vol. 10, No. 10,pp. 4498-4514.
Liu, Y. H. and Chen, Y. J. (2011), "Automatic defect detection for TFT-LCD array process using quasiconformal kernel support vector data description," International Journal of Molecular Sciences, Vol. 12, No. 9,pp. 5762-5781.
Mi, Y. (2013), "Imbalanced classification based on active learning SMOTE," Research Journal of Applied Science Engineering and Technology, Vol. 5,pp. 944-949.
Pai, P.-F., Wu, T.-M., Lin, K.-P., and Yang, S.-L. (2009), "Diagnosing faulty products in TFT-LCD manufacturing processes by support vector machines with principal components analysis," Proceedings of 2009 WRI Global Congress on Intelligent Systems.
Park, Y. and Kweon, I. S. (2016), "Ambiguous surface defect image classification of AMOLED displays in smartphones," IEEE Transactions on Industrial Informatics, Vol. 12, No. 2,pp. 597-607.
Schoen, R. R., Habetler, T. G., Kamran, F., and Bartfield, R. (1995), "Motor bearing damage detection using stator current monitoring," IEEE Transactions on industry applications, Vol. 31, No. 6,pp. 1274-1279.
Shelke, S., Thakur, A., and Pathare, Y. (2016), "Condition Monitoring of Ball Bearing Using Vibration Analysis and Feature Extraction," International Research Journal of Engineering and Technology (IRJET), Vol. 3, No. 2.
Shi, B.-C. (2011). Defect detection in TFT-LCD color filter process, Chung Hua University.
Sreejith, B., Verma, A., and Srividya, A. (2008), "Fault diagnosis of rolling element bearing using time-domain features and neural networks," Proceedings of 3rd International Conference on Industrial and Information Systems.
Tandon, N. (1994), "A comparison of some vibration parameters for the condition monitoring of rolling element bearings," Measurement, Vol. 12, No. 3,pp. 285-289.
Tsai, D. M., Chuang, S. T., and Tseng, Y. H. (2007), "One-dimensional-based automatic defect inspection of multiple patterned TFT-LCD panels using Fourier image reconstruction," International Journal of Production Research, Vol. 45, No. 6,pp. 1297-1321.
Tsai, D. M. and Hung, C. Y. (2005), "Automatic defect inspection of patterned thin film transistor-liquid crystal display (TFT-LCD) panels using one-dimensional Fourier reconstruction and wavelet decomposition," International Journal of Production Research, Vol. 43, No. 21,pp. 4589-4607.
Wang, L. and Gao, R. X. (2006), Condition monitoring and control for intelligent manufacturing. Springer Science & Business Media.
Wang, Y.-C., Lin, B.-S., and Yang, K.-H. (2014), "Flaw detection and measurement for 4K Ultra HD thin-film-transistor array panel," Measurement, Vol. 51,pp. 236-240.
Wu, A., Zhu, J., Tao, Z., and Mao, C. (2016), "Automatic inspection and classification for thin-film transistor liquid crystal display surface defects based on particle swarm optimization and one-class support vector machine," Advances in Mechanical Engineering, Vol. 8, No. 11,pp. 1687814016677660.