隨著全球暖化問題加劇，各國積極訂定2050年淨零排放目標，並針對氣候變遷提出應對策略。台灣政府於2022年公布的「2050淨零排放路徑藍圖」中，電力能源去碳化為關鍵策略之一，預計至 2050年再生能源將占總電力供應的60～70%，以太陽能與風力發電等變動性再生能源為主。然而，這些能源具有間歇性與不穩定性，對電網調度造成挑戰，因此短期電力負載預測成為提升電力調度效率的重要關鍵。
本研究旨在提昇短期電力負載預測的準確性與穩定性，針對天氣變化、季節性影響及突發性負載波動等因素，將指數加權移動平均(EWMA)方法用於電力負載數據平滑處理，並將處理後的特徵納入機器學習模型。研究比較了隨機森林(Random Forest)和極限梯度提升(XGBoost)兩種機器學習模型在「未納入EWMA特徵值」與「納入EWMA特徵值」兩種情境下的預測表現，並測試不同 EWMA 平滑參數(λ=0.1~0.5) 對預測結果的影響。此外，透過EWMA管制圖進行誤差監控，有效識別負載數據中的異常點，以驗證其在提升預測穩定性方面的應用價值。
研究結果顯示，納入EWMA特徵後，模型的 MAPE 和 RMSE 顯著降低、R²亦顯著提升，其中 λ=0.5 表現最佳，顯示適當平滑可有效減緩短期波動對模型預測的影響。考量季節分群結果顯示，各區域在不同季節的預測表現有所差異：北部區域於春、夏和冬季的預測準確度均有所提升；中部四季均提升，且夏季誤差最低；南部在冬季表現改善；東部於夏、秋、冬季均有提升。此外，基準模型(RF BASE 與 XGB BASE)的異常點數量普遍高於納入 EWMA 特徵的模型，顯示 EWMA 有助於減少異常波動。
本研究首次將 EWMA 作為機器學習特徵應用於短期電力負載預測，並以實際負載資料驗證其可行性。結果不僅顯示 EWMA 在提升短期負載預測準確與穩定性方面具有顯著效益，亦拓展了特徵處理方法在電力需求預測領域的應用潛力。

In 2022, Taiwan announced its "2050 Net-Zero Emissions Pathway Blueprint," targeting 60–70% of electricity generation from renewable sources by 2050. The intermittent nature of solar and wind generation poses serious challenges to grid stability, highlighting the need for hourly load forecasting. This paper proposes a machine learning scheme that uses Exponentially Weighted Moving Average (EWMA) features in conjunction with Random Forest (RF) or eXtreme Gradient Boosting (XGBoost) models for short-term electricity load forecasting based on seasonal trends and weather data. Model performance was evaluated in terms of mean absolute percentage error (MAPE), root mean squared error (RMSE), and R² metrics.
Our findings demonstrated that forecasting accuracy can be improved by including EWMA features, particularly during seasons with pronounced load fluctuations. EWMA control charts also proved effective in detecting forecasting anomalies, which is crucial to enhancing load prediction stability.
These findings suggest that integrating EWMA into machine learning models provides a robust and scalable strategy to improve load forecasting accuracy, supporting reliable grid operations in the context of growing renewable energy penetration.

1. 王奕夫 (2019)，利用統計製程控制與機器學習建立風力發電機之預測性維修：以彰工風力發電站為例，國立中興大學碩士論文，臺灣博碩士論文知識加值系統，https://hdl.handle.net/11296/v43g92。
2. 交通部中央氣象局，臺灣的溫度和雨量特徵全書，https://www.cwb.gov.tw/V8/C/K/Encyclopedia/climate/climate2_all.html
3. 李豫佳 (2023)，考量季節型態的短期區域電力負載機率性預測，國立成功大學碩士論文，臺灣博碩士論文知識加值系統，https://hdl.handle.net/11296/y38p64。
4. 沈家瑜 (2015)，利用智慧電表於短期電力負載預測，國立清華大學碩士論文，臺灣博碩士論文知識加值系統，https://hdl.handle.net/11296/tm6d65。
5. 林秉毅 (2018)，電力負載預測準確性之改善研究，博士論文，中原大學，臺灣博碩士論文知識加值系統，https://hdl.handle.net/11296/j335zx。
6. 洪英章 (2024)，電力峰值負載預測與 Covid-19 疫情對電力需求之區域差異分析，國立陽明交通大學博士論文，臺灣博碩士論文知識加值系統，https://hdl.handle.net/11296/pjq43b。
7. 國家發展委員會 (2022)，都市及區域發展統計彙編。
8. 國家發展委員會 (2022)，臺灣 2050 淨零排放路徑及策略總說明 (全文)。
9. 國家發展委員會 (2022)，臺灣 2050 淨零排放路徑及策略總說明 (簡報)。
10. 黃凡維 (2006)，應用 EWMA 值於類神經網路以監控製程平均，國立雲林科技大學碩士論文，臺灣博碩士論文知識加值系統，https://hdl.handle.net/11296/zayw3c。
11. 鄭伊秀、陳翔雄 (2021)，電業營運的基礎－探討國外電力負載預測做法，臺灣經濟研究月刊，44(12)，99-106。
12. 籃宏偉 (2011)，區域電力負載與氣候相關性研究－多元線性迴歸模型，國立高雄應用科技大學碩士論文，臺灣博碩士論文知識加值系統，https://hdl.handle.net/11296/e2r7e4。
13. Alfares, H. K., & Nazeeruddin, M. (2002). Electric load forecasting: Literature survey and classification of methods. International Journal of Systems Science, 33(1), 23-34.
14. Alfasanah, Z., Niam, M. Z. H., Wardiani, S., Ahsan, M., & Lee, M. H. (2025). Monitoring air quality index with EWMA and individual charts using XGBoost and SVR residuals. MethodsX, 14, 103107.
15. Bisgaard, S., & Kulahci, M. (2011). Time series analysis and forecasting by example. John Wiley & Sons.
16. Breiman, L. (2001). Random forests. Machine Learning, 45, 5-32.
17. Cao, W., Liu, Y., Mei, H., Shang, H., & Yu, Y. (2023). Short-term district power load self-prediction based on improved XGBoost model. Engineering Applications of Artificial Intelligence, 126, 106826.
18. Chen, T., & Guestrin, C. (2016, August). XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 785-794).
19. Chen, W., Rong, F., & Lin, C. (2025). Short-term building electricity load forecasting with a hybrid deep learning method. Energy and Buildings, 115342.
20. Cheng, Y. Y., Chan, P. P., & Qiu, Z. W. (2012, July). Random forest based ensemble system for short term load forecasting. In 2012 International Conference on Machine Learning and Cybernetics (Vol. 1, pp. 52-56). IEEE.
21. Dudek, G. (2015). Short-term load forecasting using random forests. In Intelligent Systems' 2014: Proceedings of the 7th IEEE International Conference Intelligent Systems IS’2014, September 24‐26, 2014, Warsaw, Poland, Volume 2: Tools, Architectures, Systems, Applications (pp. 821-828). Springer International Publishing.
22. Fan, G. F., Yu, M., Dong, S. Q., Yeh, Y. H., & Hong, W. C. (2021). Forecasting short-term electricity load using hybrid support vector regression with grey catastrophe and random forest modeling. Utilities Policy, 73, 101294.
23. Hansen, H. H., Kulahci, M., & Nielsen, B. F. (2023). Statistical process control versus deep learning for power plant condition monitoring. Computers & Chemical Engineering, 178, 108391.
24. Harikrishnan, G. R., & Sreedharan, S. (2025). Advanced short-term load forecasting for residential demand response: An XGBoost-ANN ensemble approach. Electric Power Systems Research, 242, 111476.
25. Hong, T., & Shahidehpour, M. (2015). Load forecasting case study. National Association of Regulatory Utility Commissioners.
26. Huang, H. H., & Huang, Y. H. (2024). A novel green learning artificial intelligence model for regional electrical load prediction. Expert Systems with Applications, 256, 124907.
27. Kadri, F., Harrou, F., Chaabane, S., Sun, Y., & Tahon, C. (2016). Seasonal ARMA-based SPC charts for anomaly detection: Application to emergency department systems. Neurocomputing, 173, 2102-2114.
28. Karoon, K., & Areepong, Y. (2025). The efficiency of the new extended EWMA control chart for detecting changes under an autoregressive model and its application. Symmetry, 17(1).
29. Kazmi, M. W., & Noor-ul-Amin, M. (2025). Integrating machine learning based EWMA control charts for multivariate process monitoring. Computers & Industrial Engineering, 204, 111131.
30. Khan, Z. A., Ullah, A., Haq, I. U., Hamdy, M., Mauro, G. M., Muhammad, K., ... & Baik, S. W. (2022). Efficient short-term electricity load forecasting for effective energy management. Sustainable Energy Technologies and Assessments, 53, 102337.
31. Nepal, B., Yamaha, M., Yokoe, A., & Yamaji, T. (2020). Electricity load forecasting using clustering and ARIMA model for energy management in buildings. Japan Architectural Review, 3(1), 62-76.
32. Noor-ul-Amin, M., Kazmi, M. W., Alkhalaf, S., Abdel-Khalek, S., & Nabi, M. (2024). Machine learning based parameter-free adaptive EWMA control chart to monitor process dispersion. Scientific Reports, 14(1), 31271.
33. Tziolis, G., Lopez-Lorente, J., Baka, M. I., Koumis, A., Livera, A., Theocharides, S., Makrides, G., & Georghiou, G. E. (2024). Direct short-term net load forecasting in renewable integrated microgrids using machine learning: A comparative assessment. Sustainable Energy, Grids and Networks, 37, 101256.
34. Xie, J., Liu, B., Lyu, X., Hong, T., & Basterfield, D. (2015, October). Combining load forecasts from independent experts. In 2015 North American Power Symposium (NAPS) (pp. 1-5). IEEE.