隨著氣候變遷與碳中和政策的全球推動，製造業碳排放管理成為實現永續發展的關鍵。特別是在高科技製造場域中，設備的能源使用與碳排放變化頻繁，傳統碳盤查方法難以即時掌握異常排放與潛在風險。為此，本研究提出一套結合平滑卷積神經網路（Smoothed Convolutional Neural Network）與注意力機制（Attention Mechanism）之碳排放預測與異常原因分析模型，建立具備即時性與準確性的智慧碳管理系統。
首先，針對設備所產生之碳排數據進行資料前處理與特徵工程，解決其高雜訊、缺失值與非平穩性等問題。接著，設計 S-CNN 模型進行時間序列預測，透過引入平滑機制降低模型對雜訊的敏感度，並強化其趨勢捕捉能力。在預測模型基礎上，本研究進一步整合注意力機制與隨機森林（Isolation Forest）等工具，建構可視化的異常解釋模組，能有效識別異常排放事件及其成因參數。
本研究運用製造場域150 天碳排放資料實驗，進行預測與異常原因分析測試，結果顯示，S-CNN 模型具備良好之預測準確度，並成功預測出月度排放變化趨勢與高風險時段。異常原因分析模組亦可清晰指出異常的關鍵變數，有效提升模型結果之可解釋性與決策參考價值。
本研究所建構之模型兼具理論創新與實務應用潛力，提供製造業一套可即時部署且具透明度之設備層級碳管理方法與技術。放關未來展望，可進一步擴展至多設備聯動分析與跨廠區碳排預測，以支援企業之智慧永續轉型。

Amid the global pursuit of carbon neutrality and intensified climate change policies, effective carbon emission management has become imperative for achieving sustainability in the manufacturing sector. In high-tech production environments, equipment-level carbon emissions exhibit frequent and rapid fluctuations, rendering traditional inventory methods inadequate for real-time monitoring and anomaly detection. To address these challenges, this study proposes an intelligent forecasting and diagnostic framework that integrates Smoothed Convolutional Neural Networks (S-CNN) with an Attention mechanism for carbon emission prediction and root cause analysis of anomalies.
The methodology includes preprocessing of noisy and non-stationary equipment-level emission data, followed by time series forecasting using S-CNN to improve trend learning and reduce noise sensitivity. An Attention-based diagnostic module, combined with Isolation Forest, is further developed to identify abnormal emission events and their key influencing factors.
Empirical results using real-world manufacturing data demonstrate that the proposed model achieves high accuracy, with a Mean Absolute Percentage Error (MAPE) of 2.4%, and successfully captures monthly emission trends and high-risk intervals. The diagnostic module enhances interpretability by revealing critical features associated with anomalies.
The study presents a robust and deployable solution for equipment-level carbon management. The proposed framework offers strong potential for practical implementation and future extension to multi-equipment and cross-site applications, supporting the transition to intelligent and sustainable manufacturing.

Adhikari, N., Li, H., & Gopalakrishnan, B. (2025). A Bibliometric and Systematic Review of Carbon Footprint Tracking in Cross-Sector Industries: Emerging Tools and Technologies. Sustainability, 17(9), 4205.
Bazigu, A., & Mwebaze, J. （2025）. A Framework for IoT Enabled Smart Manufacturing for Energy and Resource Optimization. arXiv preprint.
Chen, T., & Guestrin, C. （2016）. XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785–794.
Ezenkwu, C. P., Cannon, S., & Ibeke, E. （2024）. Monitoring carbon emissions using deep learning and statistical process control: A strategy for impact assessment of governments’ carbon reduction policies. Environmental Monitoring and Assessment, 196（3）, Article 231.
Franco de la Peña, M., Perales Gómez, Á. L., & Fernández Maimó, L. （2025, June 2）. ShaTS: A Shapley based Explainability Method for Time Series Artificial Intelligence Models applied to Anomaly Detection in Industrial Internet of Things [Preprint]. arXiv.
Guo, Z., Yuan, X., Zhou, K., Fu, L., & Song, Y. （2024）. How Does the Digital Transformation Affect the Carbon Emissions of Manufacturing Enterprises in China? The Perspective of Green Technology Innovation. Sustainability, 16（8）, 3184.
Han, Z., Cui, B., Xu, L., Wang, J., & Guo, Z. (2023). Coupling LSTM and CNN Neural Networks for Accurate Carbon Emission Prediction in 30 Chinese Provinces. Sustainability, 15(18), 13934.
Hochreiter, S., & Schmidhuber, J. （1997）. Long short-term memory. Neural Computation, 9（8）, 1735–1780.
Huang, X., Chen, N., Deng, Z. & Huang, S.（2024）。Multivariate time series anomaly detection via dynamic graph Attention network and Informer。Applied Intelligence, 54，7636–7658.
IPCC. （2021）. Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press.
Jiang, Y. & Mao, Z. （2024）. A novel carbon emission monitoring method for power generation enterprises based on hybrid transformer model. Scientific Reports.
Jin, W., Hong, Z., & Long, H. （2024）. Digital transformation empowers ESG performance in the manufacturing industry: From ESG to DESG. SAGE Open, 14（1）, 215824402412306.
Kaymakci, C., Wenninger, S., & Sauer, A. (2021). Energy anomaly detection in industrial applications with long short-term memory-based autoencoders. Procedia CIRP, 104, 182–187.
LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11), 2278–2324.
Li, X., Li, M., Li, H., Li, X., Du, Z., & Deng, Y. （2021）. Carbon emission prediction based on a hybrid CNN-LSTM model. Applied Sciences, 11（3）, 1252.
Li, X., Wang, B., Peng, T., & Xu, X. （2021）. Greentelligence: Smart Manufacturing for a Greener Future. Chin. J. Mech. Eng. 34, 116.
Moosavi, S., Farajzadeh-Zanjani, M., Razavi-Far, R., Palade, V., & Saif, M. （2024）. Explainable AI in Manufacturing and Industrial Cyber–Physical Systems: A Survey. Electronics, 13（17）, 3497
Moreno, S. et al. （2024）. An Ensemble Method for Non Intrusive Load Monitoring （NILM） Applied to Deep Learning Approaches. Energies, 17（18）, 4548.
Nguyen, T. T., Nguyen, T. T., Nguyen, T. D., & Pham, V. H. （2020）. An Attention-based LSTM model for carbon emission prediction. IEEE Access, 8, 150661–150673.
Patro, S. G. K., & Sahu, K. K. (2015). Normalization: A preprocessing stage. International Advanced Research Journal in Science, Engineering and Technology (IARJSET), 2(3), 20–22.
Smyl, S. （2020）. A hybrid method of exponential smoothing and recurrent neural networks for time series forecasting. International Journal of Forecasting, 36（1）, 75–85.
Tealab, A., Hefny, H., & Badr, A. (2017). Forecasting of nonlinear time series using ANN. Future Computing and Informatics Journal, 2(1), 39–47.
Tritscher, J., Krause, A., & Hotho, A. （2023）. Feature relevance XAI in anomaly detection: Reviewing approaches and challenges. Frontiers in Artificial Intelligence, 6.
United Nations. （2015）. Paris Agreement. United Nations Framework Convention on Climate Change （UNFCCC）.
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., ... & Polosukhin, I. （2017）. Attention is all you need. Advances in Neural Information Processing Systems, 30.
Wibawa, A.P., Utama, A.B.P., Elmunsyah, H. et al. （2022）. Time-series analysis with Smoothed Convolutional Neural Network. Journal of Big Data, 9（1）, 44.
World Economic Forum. （2024）. The Net-Zero Industry Tracker.
World Resources Institute (WRI) & World Business Council for Sustainable Development (WBCSD). (2004). The Greenhouse Gas Protocol: A Corporate Accounting and Reporting Standard (Revised Edition).
Xia, X., Zhu, D., Sha, J., Ma, R., & Kang, W. （2024）. Research on industrial carbon emission prediction method based on CNN–LSTM under dual carbon goals. International Journal of Low Carbon Technologies.
Yang, H. （2024）. Study on real-time estimation of carbon emissions based on non-intrusive power load monitoring. Applied and Computational Engineering, 56, 146–152.
Ye, J., & Xu, W. （2023）. Carbon reduction effect of ESG: Empirical evidence from listed manufacturing companies in China. Frontiers in Ecology and Evolution, 11, 1311777.
Yin, Y., & Yang, Y. （2025）. Sustainable transition of the global semiconductor industry: Challenges, strategies, and future directions. Sustainability, 17（7）, 3160.
Zamanzadeh Darban, Z., Webb, G. I., Pan, S., Aggarwal, C. C., & Salehi, M. (2024). Deep Learning for Time Series Anomaly Detection: A Survey. ACM Computing Surveys, 57(1), 1–42.
Zhang, L., Su, H., Zio, E., Zhang, Z., Chi, L., Fan, L., Zhou, J., & Jiang, L.（2021）. A data driven approach to anomaly detection and vulnerability dynamic analysis for large scale integrated energy systems. Energy Conversion and Management, 234, 113926.
Zhou, Z., Ma, Z., & Lin, X. （2022）. Carbon disclosure, financial transparency and agency cost: Evidence from Chinese manufacturing listed companies. Journal of Cleaner Production, 330, 129888.