研究蒐集晶圓廠實際運轉數據，整合廠務與製造部門所提供之週次資料，包括用電紀錄、設備稼動狀況與環境負載條件。透過統計分析與迴歸方法進行變數篩選與模型建立，以平均週需量作為主要預測目標。模型驗證結果顯示，能有效掌握負載高峰變化，並可進一步推估最大需量，具備良好準確度與穩定性。
經 2024 年全年 52 週回測驗證，模型週平均需量預測之平均絕對百分比誤差（MAPE）為 0.90%，誤差範圍介於 -3.2% 至 +2.1% 之間，顯示模型於建模資料上具有良好擬合能力。研究成果能協助晶圓廠更精準地預估未來用電趨勢，據以調整契約容量，避免不必要的基本電費支出，並降低因超約導致之罰款風險。整體而言，研究提供一套實用的決策工具，能有效支持企業在營運成本控管與能源效率提升上的需求，對半導體產業的競爭力具實質助益。
研究並以 2025 年全年（第 1 至第 53 週）之非建模期間資料進行外部驗證，模型週平均需量預測之 MAPE 為 1.10%，偏差百分比介於 -1.44% 至 4.96% 之間，確認模型具備良好之跨年度泛化能力。實務效益分析進一步顯示，與 ρᵢ 契約容量估算法相比，本研究方法於 2025 年全年可降低年度總電費約 389,100 元，約占 ρᵢ 契約容量估算法年度總費用之 0.56%，且各月均未發生超約，具體量化研究模型於成本最佳化與風險控管上之實務貢獻。

This research developed a mid-term demand forecasting model to address the high energy consumption and continuous operation of wafer fabrication plants in the Southern Taiwan Science Park. As contract capacity directly affects basic electricity charges and excess penalties, accurate peak demand forecasting is vital for efficiency management.
Raw 15-minute electricity consumption data from actual fab operations were aggregated into weekly datasets for modeling, then combined with equipment utilization and environmental load conditions. Statistical and regression methods were applied to identify significant variables and build the model, with average weekly demand as the prediction target. Validation confirmed the model effectively captures peak demand variations and provides reliable maximum demand estimates with strong accuracy and stability.
Year-long 2024 backtesting demonstrated strong in-sample accuracy (MAPE = 0.90%, weekly deviation from -3.2% to +2.1%) and showed that the model enables fabs to better anticipate demand and adjust contract capacity, reducing unnecessary basic charges and penalty risks. External validation using full-year 2025 data confirmed the model's generalizability (MAPE = 1.10%, weekly deviation from -1.44% to 4.96%). Benefit analysis further showed that, compared with the ρᵢ contract capacity estimation method, the proposed method reduced annual electricity costs by approximately NT$389,100 (about 0.56% of the ρᵢ contract capacity estimation method cost), while avoiding excess-demand penalties in all months.

[1] R. J. Hyndman and G. Athanasopoulos, Forecasting: Principles and Practice, 3rd ed., OTexts, 2021.
[2] Technopolis Ltd, "Evaluation of the Capacity Market scheme," Department for Energy Security and Net Zero, UK Government, 2023.
[3] 台灣電力公司，「營業規章第一章總則第二節名詞釋義」，2022 年 9 月。
[4] 台灣電力公司，「營業規章第七章〈收費〉第63條、第71條及線路設置費收費費率表」，2022 年 9 月。
[5] 國立陽明交通大學總務處，「節能減碳」，國立陽明交通大學，2023 年，取自 https://www.nycu.edu.tw/ga/ch/app/artwebsite/view?module=artwebsite&id=21326&serno=10514a2f-87ca-4f76-86c6-e7f1c8728b24
[6] 台灣電力公司，「契約容量申請與變更說明」，台灣電力公司，2024 年。
[7] 台灣電力公司，「營業規章第六章〈計費〉第67條、第68條」，2022 年 9 月。
[8] 台灣電力公司，「各類電價表及計算範例」，台灣電力公司，2024 年。
[9] Y. A. Çengel and A. J. Ghajar, Heat and Mass Transfer: Fundamentals and Applications, 5th ed., McGraw-Hill, 2015.
[10] ASHRAE, ASHRAE Handbook—Fundamentals, Chapter 1: Psychrometrics, American Society of Heating, Refrigerating and Air-Conditioning Engineers, 2021.
[11] S.-C. Hu and J.-M. Tsao, "A comparative study on energy consumption for HVAC systems of high-tech FABs," Appl. Therm. Eng., vol. 27, no. 17–18, pp. 2758–2766, 2007.
[12] S.-C. Hu and Y. K. Chuah, "Power consumption of semiconductor fabs in Taiwan," Energy, vol. 28, no. 8, pp. 895–907, 2003.
[13] J.-M. Tsao, S.-C. Hu, T. Xu, and D. Y.-L. Chan, "Capturing energy-saving opportunities in make-up air systems for cleanrooms of high-technology fabrication plant in subtropical climate," Energy Build., vol. 42, no. 11, pp. 2005–2013, 2010.
[14] Y. Lin, T. Huang, W. Yang, X. Hu, and C. Li, "A review on the impact of outdoor environment on indoor thermal environment," Buildings, vol. 13, no. 10, Art. no. 2600, 2023.
[15] K. Kircher, X. Shi, S. Patil, and K. M. Zhang, "Cleanroom energy efficiency strategies: Modeling and simulation," Energy Build., vol. 42, no. 3, pp. 282–289, 2010.
[16] K. Chen, K. Chen, Q. Wang, Z. He, J. Hu, and J. He, "Short-term load forecasting with deep residual networks," IEEE Trans. Smart Grid, vol. 10, no. 4, pp. 3943–3952, 2019.
[17] N. Amjady and F. Keynia, "Mid-term load forecasting of power systems by a new prediction method," Energy Convers. Manage., vol. 49, no. 10, pp. 2678–2687, 2008.
[18] S. S. W. Fatima and A. Rahimi, "A review of time-series forecasting algorithms for industrial manufacturing systems," Machines, vol. 12, no. 6, Art. no. 380, 2024.
[19] D. C. Montgomery, E. A. Peck, and G. G. Vining, Introduction to Linear Regression Analysis, 5th ed., John Wiley & Sons, 2012.
[20] N. R. Draper and H. Smith, Applied Regression Analysis, 3rd ed., John Wiley & Sons, 1998.
[21] L. Breiman, "Random forests," Mach. Learn., vol. 45, no. 1, pp. 5–32, 2001.
[22] G. James, D. Witten, T. Hastie, and R. Tibshirani, An Introduction to Statistical Learning: With Applications in R, 2nd ed., Springer, 2021.
[23] T. Hastie, R. Tibshirani, and J. Friedman, The Elements of Statistical Learning: Data Mining, Inference, and Prediction, 2nd ed., Springer, 2009.
[24] A. Liaw and M. Wiener, "Classification and regression by randomForest," R News, vol. 2, no. 3, pp. 18–22, 2002.
[25] M. Kuhn and K. Johnson, Feature Engineering and Selection: A Practical Approach for Predictive Models, CRC Press, 2019.
[26] Y. Saeys, I. Inza, and P. Larrañaga, "A review of feature selection techniques in bioinformatics," Bioinformatics, vol. 23, no. 19, pp. 2507–2517, 2007.
[27] M. Kuhn and K. Johnson, Applied Predictive Modeling, Springer, 2013.
[28] 李翔祐，「電力負載預測及契約容量最佳化之研究」，國立臺北科技大學電機工程系碩士論文，2021 年。
[29] S.-H. Tai, M.-T. Tsai, W.-H. Huang, and Y.-H. Tsai, "Contracted capacity optimization problem of industrial customers with risk assessment," Inventions, vol. 9, no. 4, Art. no. 81, 2024.
[30] M. T. Tsay, W. M. Lin, and J. L. Lee, "Optimal contracts decision of industrial customers," Int. J. Electr. Power Energy Syst., vol. 23, no. 8, pp. 795–803, 2001.
[31] W. Cai, F. Liu, H. Zhang, P. Liu, and J. Tuo, "Development of dynamic energy benchmark for mass production in machining systems for energy management and energy-efficiency improvement," Appl. Energy, vol. 202, pp. 715–725, 2017.
[32] J. M. Juran and F. M. Gryna, Juran's Quality Control Handbook, 4th ed., McGraw-Hill, 1988.
[33] International Organization for Standardization, ISO 50001:2018 Energy Management Systems—Requirements with Guidance for Use, 2018.
[34] International Organization for Standardization, ISO 50006:2014 Energy Management Systems—Measuring Energy Performance Using Energy Baselines (EnB) and Energy Performance Indicators (EnPI)—General Principles and Guidance, 2014.