隨著電子產品朝向高密度化與微型化發展，表面黏著技術製程之品質穩定性已成為電子製造業的重要議題，其中，錫膏印刷檢測與自動光學檢測分別扮演前段製程監控與後段品質驗證的重要角色，然而，目前多數SMT製程仍以事後檢驗與傳統統計製程管制方式進行品質管理，缺乏利用大量製程資料進行即時預測與異常預警之能力，容易造成不良品流出與重工成本增加。
本研究以某SMT生產線之SPI與AOI製程資料為研究對象，共蒐集八十三萬多筆焊點資料，並以錫膏體積、錫膏高度、缺錫情形、X軸偏移量、Y軸偏移量與SPI良率等製程參數作為輸入變數，以AOI檢測結果作為分類目標，研究中利用邏輯斯迴歸與決策樹建立缺陷預測模型，並導入成本敏感學習處理資料不平衡問題；同時結合p-CUSUM統計製程管制方法，建立AOI良率之即時監控機制，以提升製程異常預警能力。
研究結果顯示，邏輯斯迴歸模型之整體分類正確率達九成四，AUC值接近1；決策樹模型之整體分類正確率則達九成八，且具備良好之規則解釋能力，此外，p-CUSUM管制圖能有效偵測AOI良率之微小偏移，顯示其適用於高良率SMT製程監控，研究結果證明，結合資料探勘與統計製程管制方法，可有效提升SMT製程品質監控與缺陷預測能力，並提供電子製造業導入智慧化品質管理之參考。

Surface Mount Technology (SMT) manufacturing generates substantial inspection data through Solder Paste Inspection (SPI) and Automated Optical Inspection (AOI) systems. These data are often utilized only for post-process inspection rather than predictive quality management. This study proposes a framework integrating classification algorithms and statistical process monitoring to improve defect prediction and manufacturing quality control. A dataset containing 839,616 solder-joint records was collected from a SMT production line. Logistic regression and decision tree learning were employed to predict AOI inspection outcomes using SPI process parameters. Furthermore, a Cumulative Sum (CUSUM) control chart was implemented to monitor AOI yield rates and detect process shifts. The experimental results showed that both logistic regression and decision tree models achieved high prediction accuracies, and that SPI yield rate and the volume and height of solder paste were the most influential variables of AOI outcomes. The CUSUM monitoring scheme effectively detected process variation trends and provided early warning signals. The proposed framework effectively integrates classification models and monitoring functions into a unified quality management system for SMT manufacturing.

Aaboub, F., Chamlal, H., & Ouaderhman, T. (2023). Statistical analysis of various splitting criteria for decision trees. Proceedings of the Institution of Mechanical Engineers, Part I: Journal of Systems and Control Engineering, 237(10), 2062–2076.
Abd Al Rahman, M., & Mousavi, A. (2020). A review and analysis of automatic optical inspection and quality monitoring methods in electronics industry. IEEE Access, 8, 183192–183271.
Ahn, J., Kim, H., Park, J., & Lee, S. (2024). A comprehensive survey on machine learning driven material defect detection. Journal of Manufacturing Systems, 78, 1–25.
Cao, N., Farrag, A., Won, D., & Yoon, S. W. (2024). XSCAN: Explainable solder joint defect probability prediction through solder paste printing status with imbalanced data. Journal of Manufacturing Systems, 77, 212–227.
Cerquitelli, T., Chiusano, S., & Garza, P. (2021). Manufacturing as a data-driven practice: Methodologies, tools and challenges. IEEE Access, 9, 29869–29892.
Eisentraut, L., Hosch, J., Roytenberg, M., Benecke, A., Penava, P., & Buettner, R. (2025). Defect detection in industrial soldering processes using machine learning: A critical literature review. IEEE Access, 13, 41533–41558.
Fries, N., Müller, T., & Kampmann, A. (2024). Data-driven process adjustment policies for quality improvement in manufacturing. Expert Systems with Applications, 244, 122963.
Gee, N., May, M., & Kiefer, L. (2022). Smart manufacturing in stencil printing operations. Journal of Manufacturing Systems, 48, 144–156.
Ghahramani, M., Qiao, Y., Zhou, M. C., & Zhang, Y. (2021). Data-driven predictive analysis for smart manufacturing: A survey. IEEE Transactions on Automation Science and Engineering, 18(2), 623–637.
Gunraj, H., Wuthrich, M., & Wong, A. (2023). Towards trustworthy visual inspection of solder joints in electronics manufacturing. Advanced Intelligent Systems, 5(8), 2300092.
Han, D., & Tsung, F. (2023). The optimal CUSUM control chart with a dynamic non-random control limit and a given sampling strategy for small samples sequence. Quality and Reliability Engineering International, 39(2), 567–584.
Haridy, S., & Benneyan, J. C. (2024). Shewhart–EWMA chart for monitoring binomial data subject to shifts of random amounts. Computers & Industrial Engineering, 193, 110252.
He, Z., Xie, M., & Goh, T. N. (2020). Generalized linear model based monitoring methods for high-yield processes. Quality and Reliability Engineering International, 36(5), 1570–1591.
Hosmer, D. W., Lemeshow, S., & Sturdivant, R. X. (2013). Applied logistic regression (3rd ed.). Wiley.
Hsu, C.-C., & Chuang, C.-S. (2022). Integral equation solutions for the average run length for monitoring a SARFIMAX process with underlying exponential white noise using a CUSUM control chart. PLOS ONE, 17(2), e0264283.
Jones, C. L., Abdel-Salam, G., & Mays, D. (2022). Novel Bayesian CUSUM and EWMA control charts via various loss functions for monitoring processes. Quality and Reliability Engineering International.
Khader, N., Lu, H., May, M., & Kiefer, L. (2021). Adaptive optimal control of stencil printing process using real-time solder paste volume feedback. Journal of Manufacturing Systems, 60, 692–703.
Khaliq, Q. U. A., Aslam, M., & Jun, C.-H. (2023). Combined Shewhart–EWMA and Shewhart–CUSUM monitoring schemes for time between events. Quality and Reliability Engineering International, 39(2), 474–497.
LaCasse, P. M., Otieno, W., & Maturana, F. P. (2020). Predicting contact-without-connection defects on printed circuit boards employing BGA packages: A data analytics case study. SN Applied Sciences, 2(2), 156.
Li, H., Liu, J., Wang, Y., & Zhang, T. (2022). A reliable solder joint inspection method based on a light-weight point cloud learning network. Neurocomputing, 500, 86–98.
Liu, R., van der Aalst, W. M. P., & Oostrum, R. (2022, December 1-4). Improving the performance of automated optical inspection (AOI) using machine learning [Paper presentation]. 2022 IEEE International Conference on Industrial Engineering and Engineering Management (IEEM), Kuala Lumpur, Malaysia.
Lucas, J. M., & Crosier, R. B. (1982). Fast initial response for CUSUM quality-control schemes. Technometrics, 24(3), 199–205.
Montgomery, D. C. (2022). Introduction to statistical quality control (8th ed.). Wiley.
Mao, Y., Zhang, J., Li, H., & Wang, L. (2021). Improving accuracy of automatic optical inspection with machine learning. Frontiers of Computer Science, 15(5), 1–13.
Miao, Y., Song, D., Zhang, X., & Zhang, T. (2021). Cost-sensitive Siamese network for PCB defect verification in automated optical inspection. IEEE Transactions on Instrumentation and Measurement, 70, 1–12.
Negrão, D. O., Rodrigues, L. H., & Ribeiro, J. L. D. (2022). A PRISMA-driven systematic review of data mining methods used for defects detection and classification in the manufacturing industry. Production, 32, e20210097.
Novoselov, A., Klochkov, Y., & Shishkin, Y. (2020, October 14-16). Process optimization of SnCuNi soldering material using multivariate analysis [Paper presentation]. 2020 IEEE Conference on Actual Problems of Unmanned Aerial Vehicles Developments (APUAVD), Kyiv, Ukraine.
Pan, K., Ha, J. H., Wang, H., Xu, J., & Park, S. B. (2021). An analysis of solder joint formation and self-alignment of chip capacitors. IEEE Transactions on Components, Packaging and Manufacturing Technology, 11(1), 161–168.
Raut, R. D., Mangla, S. K., Narwane, V. S., Gardas, B. B., & Narkhede, B. E. (2023). Data-driven decision making in advanced manufacturing systems. Production Planning & Control, 34(17–18), 1993–2010.
Saeed, N., Aslam, M., & Jun, C.-H. (2024). Sensitivity and reliability comparisons of EWMA mean charts under non-normal processes. Quality and Reliability Engineering International, 40(4), 1502–1519.
Saud, A. S., Shakya, S., & Neupane, B. (2021). Analysis of depth of entropy and Gini index based decision trees for predicting diabetes. Indian Journal of Computer Science, 6(6), 19–28.
Serrano, M., Gómez, A., & Martínez, J. (2025). Evaluating the performance of serial production lines using tree-based machine learning algorithms. Advanced Studies in Design and Production Systems, 12(2), 45–63.
Suri, K., Skaf, Z., & Ratchev, S. (2023). Cyber-Physical Systems as key element to Industry 4.0: Characteristics, applications and related technologies. International Journal of Computer Integrated Manufacturing, 36(10–11), 1163–1188.
Tercan, H., & Meisen, T. (2022). Machine learning and deep learning based predictive quality in manufacturing: A systematic review. Journal of Intelligent Manufacturing, 33(8), 1903–1932.
Wang, K., Xu, W., & Li, J. (2024). A unified statistical monitoring framework for multivariate autocorrelated processes. Computers & Industrial Engineering, 198, 110675.
Wang, X., Li, Y., & Chen, Z. (2021). Splitting choice and computational complexity analysis of decision trees. Journal of Applied Mathematics and Computer Science, 31(3), 451–468.
Wu, S., & Yang, H. (2024). Optimization designs of the combined Shewhart–CUSUM control charts. Journal of Statistical Computation and Simulation, 94(5), 874–897.
Zhang, B., Sang, H., & others. (2025). Integrated heterogeneous graph and reinforcement learning enabled efficient scheduling for surface mount technology workshop. Information Sciences.
Zhang, L., Wang, H., & Li, Y. (2023). A novel contrastive self-supervised learning framework for solder joint defect detection under class imbalance. IEEE Transactions on Instrumentation and Measurement, 72, 1–13.
Zhang, Q., Liu, H., & Sun, Y. (2025). AI-enabled defect detection in industrial products: A comprehensive survey, key insights and future research challenges. Computers & Industrial Engineering, 195, 109990.
Zhu, W., Li, X., & Sun, Y. (2025). Variable selection in high-dimensional logistic regression via adaptive penalization. IEEE/ACM Transactions on Computational Biology and Bioinformatics, 22(2), 345–357.
Zhang, X., Pan, K., Ha, J. H., Wang, H., & Park, S. B. (2024). Effect of differently shaped solder joints of chip resistor on self-alignment and failure behavior. Journal of Electronic Packaging, 146(2), 021002.
Zhang, Y., Li, C., & Wang, S. (2024). Explainable machine learning models for defects detection and process improvement in manufacturing. Computers & Industrial Engineering, 195, 109870.