使用精準的負載預測在機組排程、電網安全分析及經濟調度，不僅可以達到降低運轉所需成本的目標，也可以使電力能穩定且可靠的供應。未來大量的再生能源併入電力系統中，為穩定系統運轉頻率及電壓，負載預測的準確性在未來會更加重要。本研究根據台電的歷史負載資料、溫度資料(體感溫度)以及中央氣象局的溫度預測資料(體感溫度)，分析以不同的訓練集、輸出組合，對數天的短期負載預測建立類神經網路(Artificial Neural Network, ANN)模型，考慮特殊日以及一般日的關係並建立模型，並與SARIMA(Seasonal Autoregressive Integrated Moving Average)模型及半參數可加模型(Semiparametric Additive Model, SAM)進行比較。本研究利用自動化介面，並且於每日早上九點自動擷取所需的負載資料、體感溫度資料，重新訓練模型並預測當天至未來七天負載。本論文分析每日預測誤差以驗證模型準確性；測試結果顯示，以類神經網路模型及半參數可加模型混合的短期負載預測方法提供比其他測量方法更為準確的結果。

Accurate Short-Term Load Forecast (STLF) is important in the security constrained unit commitment and economic dispatch to achieve reliable, stable and efficient power system operations. The requirement for accurate STLF is becoming greater due to the integrations of large scale variable generations in order to maintain stable frequency and voltage in future power system operations. Based on Taiwan power system historical load data and apparent temperature data, different ANN models are developed to conduct STLF. For special and normal days. Forecast results are compared with those of the Seasonal Autoregressive Integrated Moving Average (SARIMA) model and the Semiparametric Additive Model (SAM). A user interface is used to collect data at 9:00, train the models, and predict loads in the forecasting to day. Mean absolute percent errors are computed to assess the effectiveness of the proposed models. Test results indicate that a mix of ANN and SAM provides better forecast results as compared to the other tested methods.

論文審定書 i
誌謝 ii
中文摘要 iii
Abstract iv
目錄 v
圖目錄 vii
表目錄 ix
第一章 緒論 1
1.1 本論文研究背景 1
1.2 文獻回顧 2
1.2.1 時間序列分析法 3
1.2.2 人工智慧技術 7
1.3 論文架構與貢獻 9
第二章 短期負載預測模型 10
2.1 資料介紹與處理 10
2.1.1 負載資料 10
2.1.2 溫度與濕度效應 12
2.1.3 特殊日及一般日 13
2.2 類神經網路模型介紹 14
2.2.1 模型架構 20
2.2.2 模型求解方法 22
2.2.3 模型修訂程序 24
2.2.4 誤差分析 33
2.3 季節性整合型自我迴歸移動平均模型 35
2.4 半參數可加模型 36
2.5 混合模型預測方法 38
2.6 特殊日模型 39
2.6.1 特殊日各自取代型態 41
2.6.2 特殊日前後一天取代型態 66
2.6.3 特殊日負載預測程序 69
第三章 負載預測結果比較 71
3.1 類神經網路模型預測結果 72
3.2 SARIMA模型預測結果 84
3.3 半參數可加模型預測結果 85
3.4 混合模型預測結果 88
3.5 特殊日模型預測結果 91
3.6 預測結果比較與討論 92
第四章 結論與未來研究建議 94
4.1 結論 94
4.2 未來研究方向 95
參考文獻 96
附錄 98

[1] A. K, Singh, S. Khatoon, M. Muazzam, and D. K. Chaturvedi,“ Load forecasting techniques and methodologies: A review. In Power,” Control and Embedded Systems (ICPCES), 2012 2nd International Conference on IEEE, pp. 1-10, December, 2012.
[2] G. Juberias, R. Yunta, J. G. Moreno, and C. Mendivil,“ A new ARIMA model for hourly load forecasting,” In Transmission and Distribution Conference on IEEE, Vol. 1, pp. 314-319, April, 1999.
[3] R. S. Tsay, “Analysis of financial time series,” pp. 543, 2005.
[4] 林信成，軟性計算理論與知識處理技術之研究，教育資料與圖書館學，2000。
[5] T. Senjyu, S. Higa, and K. Uezato,” Future load curve shaping based on similarity using fuzzy logic approach,” IEE Proceedings-Generation, Transmission and Distribution, vol. 145, no. 4, pp. 375-380, 1998.
[6] K. H. Kim, J. K. Park, K. J. Hwang, and S. H. Kim, “Implementation of hybrid short-term load forecasting system using artificial neural networks and fuzzy expert systems,” IEEE Transactions on Power Systems, vol. 10, no. 3, pp. 1534-1539, 1995.
[7] S. I. Vagropoulos, E. G. Kardakos, C. K. Simoglou, A. G. Bakirtzis, and J. P. Catalão,“ Artificial neural network-based methodology for short-term electric load scenario generation,” In Intelligent System Application to Power Systems (ISAP), 2015 18th International Conference on IEEE, pp. 1-6, September, 2015.
[8] H. Daneshi, and A. Daneshi,” Real time load forecast in power system,” In Electric Utility Deregulation and Restructuring and Power Technologies, Third International Conference on IEEE, pp. 689-695, April, 2008.
[9] Available online: http://www.taipower.com.tw/content/new_info/new_info_in.aspx?LinkID=11
[10] R. G. Steadman,” A universal scale of apparent temperature,” Journal of Climate and Applied Meteorology, vol. 23, no. 12, pp. 1674-1687, 1984.
[11] Available online: http://e-service.cwb.gov.tw/HistoryDataQuery/
[12] 羅華強，「類神經網路: MATLAB 的應用」，高立圖書有限公司，2011。
[13] J. De Villiers, and E. Barnard,” Backpropagation neural nets with one and two hidden layers,” IEEE Transactions on Neural Networks, vol. 4, no.1, pp. 136-141, 1993.
[14] H. S. Hippert, C. E. Pedreira, and R. C. Souza,“ Neural networks for short-term load forecasting: A review and evaluation,” IEEE Transactions on power systems, vol. 16, no. 1, pp. 44-55, 2001.
[15] J. W. Taylor,“ Short-term electricity demand forecasting using double seasonal exponential smoothing,” Journal of the Operational Research Society, vol. 54, no. 8, pp. 799-805, 2003.
[16] S. Fan, and R. J. Hyndman,” Short-term load forecasting based on a semi-parametric additive model,” IEEE Transactions on Power Systems, vol. 27, no. 1, pp. 134-141, 2012.
[17] 徐朮，「電力負載量之短期預測」，中山大學應用數學系研究所學位論文， 頁碼1-35，2016。
[18] 袁子倫、徐朮、楊亞叡、董道廷、羅夢娜、吳進忠、盧展南，「電力負載量之短期預測」，中華民國第三十七屆電力工程研討會，12月，2017。
[19] Available online: http://www.moi.gov.tw/dca/01news_001.aspx?sn=8498
[20] 盧展南，「強化現行輔助服務估算機制及作業平台」，台灣電力股份有限公司委託研究案，第二次期中報告，2017年。