快速找出 TFT-LCD 製程中的異常機台，是確保產品品質的關鍵。然而，傳統統計模型在處理大規模時間序列數據時能力有限，難以從陣列製程缺陷密度數據中精準識別異常設備。本研究收集並預處理 A 公司 10000 筆數據，將數值特徵標準化，類別特徵經 One-Hot 編碼處理。隨後訓練一個 LSTM 模型，擷取長時間特徵關係的優勢，探索製程缺陷密度趨勢的關聯模式，並判斷哪些缺陷密度的趨勢是異常的。實驗結果顯示，該模型在測試集上達到 0.894 的準確率、0.859 的精確率、0.874 的召回率，ROC 曲線下面積高達 0.960，展現出卓越的異常趨勢識別能力。真實案例分析證實，該方法能夠有效辨別異常的缺陷密度趨勢，本研究闡明了 LSTM 在智慧製造領域的廣闊應用前景，為製造業數位化轉型貢獻了有力的技術支持。

Rapid identification of abnormal machines in the TFT-LCD manufacturing process is crucial for ensuring product quality. However, traditional statistical models have limited capabilities when dealing with large-scale time series data. This study aims to accurately identify anomalous equipment from array process defect density data. 10,000 defect density data points were collected and preprocessed from Company A, with numerical features standardized and categorical features processed through One-Hot encoding. Subsequently, an LSTM model was trained, leveraging its advantage in capturing long-term dependencies, to explore the association patterns between process parameters and defects and determine which defect density trends are abnormal. Experimental results show that the model achieved an accuracy of 0.894, a precision of 0.859, a recall of 0.874, and an area under the ROC curve of 0.960 on the test set, demonstrating excellent anomaly trend identification capabilities. Real-world case studies confirm that in most cases, this method can effectively distinguish abnormal defect density trends. This research highlights the broad application prospects of LSTM in the field of intelligent manufacturing and contributes powerful technical support for the digital transformation of the manufacturing industry.

Abbasimehr, H., & Paki, R. (2022). Improving time series forecasting using lstm and attention models. Journal of Ambient Intelligence and Humanized Computing, 13, 673-691.

Bansal, M., Goyal, A., & Choudhary, A. (2022). A comparative analysis of k-nearest neighbor, genetic, support vector machine, decision tree, and long short term memory algorithms in machine learning. Decision Analytics Journal, 3, 100071.

Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5-32.

Chen, L.F., Su, C.T., & Chen, M.H. (2009). A neural-network approach for defect recognition in tft-lcd photolithography process. IEEE Transactions on Electronics Packaging Manufacturing, 32(1), 1-8.

Chien, C.-F., Ling, Y.-M., Kao, S.-X., & Lin, C.-H. (2022). Image-based defect classification for TFT-LCD array via convolutional neural network. IEEE Transactions on Semiconductor Manufacturing, 35(4), 650-657.

Cortes, C., & Vapnik, V. (1995). Support-vector networks. Machine Learning, 20(3), 273-297.

Cui, J., Cao, X., Zhao, F., Zhong, Z., & Han, N. (2021). Comparison of the melting properties of glass substrates for thin film transistor liquid crystal display and organic light-emitting diode. Journal of Asian Ceramic Societies, 9(3), 902-915.

Dutta, V., Choraś, M., Pawlicki, M., & Kozik, R. (2020). A deep learning ensemble for network anomaly and cyber-attack detection. Sensors, 20(16), 4583.

Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2014). Generative adversarial nets. In Advances in Neural Information Processing Systems, 27(1), 139-144.

Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735-1780.

Katayama, M. (1999). Tft-lcd technology. Thin Solid Films, 341(1-2), 140-147.

Kim, E., Cho, S., Lee, B., & Cho, M. (2019). Fault detection and diagnosis using self-attentive convolutional neural networks for variable-length sensor data in semiconductor manufacturing. IEEE Transactions on Semiconductor Manufacturing, 32(3), 302–309.

Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. Advances in Neural Information Processing Systems, 25(1), 1097-1105.

LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436-444.

Liu, Y. H., & Chen, Y. J. (2011). Automatic defect detection for TFT-LCD array process using quasiconformal kernel support vector data description. International Journal of Molecular Sciences, 12(9), 5762-5781.

Lu, H., & Ma, X. (2020). Hybrid decision tree-based machine learning models for short-term water quality prediction. Chemosphere, 249, Article 126169.

Medsker, L. R., & Jain, L. C. (2001). Recurrent neural networks. Design and Applications, 5(64-67), 2.

Mulyani, R., Sari, Y. P., & Sumantriyadi, S. (2022). Forecasting produksi perikanan budidaya di kota palembang dengan metode autoregressive integrated moving average. Sainmatika: Jurnal Ilmiah Matematika Dan Ilmu Pengetahuan Alam, 19(2), 163-174.

Murugesan, R., Mishra, E., & Krishnan, A. H. (2022). Forecasting agricultural commodities prices using deep learning-based models: Basic LSTM, bi-LSTM, stacked LSTM, CNN LSTM, and convolutional LSTM. INTERNATIONAL JOURNAL OF SUSTAINABLE AGRICULTURAL MANAGEMENT AND INFORMATICS, 8(3), 242–277. https://doi.org/10.1504/IJSAMI.2022.125757
O’Mara, W. C. (1993). Liquid Crystal Flat Panel Displays: Manufacturing Science & Technology. Van Nostrand Reinhold, New York.

Panjan, P., Drnovšek, A., Gselman, P., Čekada, M., & Panjan, M. (2020). Review of growth defects in thin films prepared by pvd techniques. Coatings, 10(5), 447.

Park, B. C., Park, E. S., Choi, B. K., Kim, B. H., and Lee, J. H. (2008), Simulation based planning and scheduling system for TFT-LCD fab, Proceedings of the 40th Conference on Winter Simulation, 2271-2276. New York: IEEE.

Pham, T., Tran, T., Phung, D., & Venkatesh, S. (2017). Predicting healthcare trajectories from medical records: a deep learning approach. Journal of Biomedical Informatics, 69, 218-229.

Quinlan, J. R. (1986). Induction of decision trees. Machine Learning, 1(1), 81-106.

Rahman, L., Mohammed, N., & Al Azad, A. K. (2016). A new LSTM model by introducing biological cell state. 2016 3rd International Conference on Electrical Engineering and Information Communication Technology, 1-6. haka, Bangladesh.

Shariff, S. M. (2022). Autoregressive integrated moving average and Long Short-Term Memory network models for forecasting energy consumptions. European Journal of Electrical Engineering and Computer Science, 6(3), 7-10.

Shin, H. J., & Kang, Y. H. (2010). A rework-based dispatching algorithm for module process in tft-lcd manufacture. International Journal of Production Research, 48(3), 915-931.

Sivapalan, G., Nundy, K. K., Dev, S., Cardiff, B., & John, D. (2022). A lightweight neural network for ecg anomaly detection in iot edge sensors. IEEE Transactions on Biomedical Circuits and Systems, 16(1), 24-35.

Troutman, R. R., Jenkins, L. C., Polastre, R. J., & Wisnieff, R. L. (1991). Characterization of TFT/LCD arrays. Conference Record of the 1991 International Display Research Conference, 231-234. Hyatt Islandia, San Diego, California.

Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017). Attention is all you need. 31st Conference on Neural Information Processing Systems, 30. Long Beach, CA, USA.

Verma, D., & Kumar, S. (2017). Prediction of defect density for open source software using repository metrics. Journal of Web Engineering, 61(4), 293-310.

Wu, D., Jiang, Z., Xie, X., Wei, X., Yu, W., & Li, R. (2020). LSTM learning with bayesian and gaussian processing for anomaly detection in industrial IoT. IEEE Transactions on Industrial Informatics, 16(8), 5244-5253.