許多電力系統的操作都仰賴於負載預測，包括發電機組排程、安全電網分析等等，代表 負載預測是電力系統有效運行與規劃的重要工作。電力需求的高估將會造成過於保守的運轉，導致啟動許多機組去供應非必要的備轉容量或過多的能源採購以及在電力設施的大量投資浪費；另一方面，低估可能會導致供應無法達到需求量及帶來運轉上風險，並造成即時備轉容量的準備不足，使電力系統暴露在較為危險的運轉範圍內操作。

本研究將依據系統歷史負載資料，氣象歷史資料 (體感溫度)，氣象局氣象預測資料 (體感溫度) 來進行負載預估，並考慮負載序列具有日內時段週期及週內週期兩種循環，建立半參數迴歸模型，且在模型的殘差考慮時間序列效應。評估模型的依據為，預測未來 24 小時的負載量，與真實值比較計算誤差。在建立模型的過程中，考慮有特殊假日的影響 (農曆春節與國定假日)，又可分成不同的處理方式。利用預測日的前三週作為訓練樣本，計算出在 2012 年至 2015 年的平均絕對比例誤差，以呈現此短期負載預測方法的有效性。

Short-term load forecasting is an essential instrument in power system planning, operation, and control. Many operations of power system are based on load forecasting, such as unit commitment problem, network security analysis etc. Overestimation of electricity load requirement will cause conservative operations, and result in many unnecessary transfers of electrical supplies or excessive waste of energy procurement, as well as costly large investments in electrical utilities. On the other hand, underestimation will cause the power generation not meeting the electricity demand, and real-time transfer capacity not meeting the safe operating reserve which may put the power system in risk.

In this work, short-term load forecast is made with historical loads, corresponding apparent temperatures and predicted apparent temperatures from Center Weather Bureau. Semi-parameter additive model is built with intra-daily and intra-weekly effect. Time-series model is also considered for the residual effects. To test the model, the future 24 hours' electricity load forecasts are made and compared to the real demands obtained later on. The special days including the Chinese New Year and National holidays are dealt with differently. The mean absolute percentage errors (MAPE) for the last four years (2012-2015) on the load forecasts, based on three weeks training data before the forecasting date, are provided to demonstrate the effectiveness of the current short-term load forecasting methodology.

[論文審定書 i]
[誌謝 ii]
[摘要 iii]
[Abstract iv]
[1 前言 1]
[2 資料介紹與處理 2]
[3 短期預測模型 4]
[3.1 日期效應 4]
[3.2 溫度與濕度效應 7]
[3.3 誤差項處理 9]
[3.4 短期預測 11]
[3.5 混和型方法 13]
[3.6 特殊假日分析 14]
[4 模型修正過程 15]
[5 預測結果 17]
[5.1 一般日預測結果 17]
[5.2 特殊假日預測結果 19]
[5.3 混和型方法預測結果 20]
[6 結論與建議 21]
[7 參考文獻 22]
[8 附錄 23]
[8.1 附錄一 23]
[8.2 附錄二 25]

1. Cho, H., Goude, Y., Brossat, X., & Yao, Q. (2013). Modeling and forecasting daily electricity load curves: a hybrid approach. Journal of the American Statistical Association, 108, 7-21.
2. Cho, M. Y., Hwang, J. C., & Chen, C. S. (1995). Customer short term load forecasting by using ARIMA transfer function model. In Energy Management and Power Delivery. Proceedings of EMPD'95., 1995 International Conference. IEEE.
3. Cryer, J. D., & Kellet, N. (1986). Time series analysis. Boston: Duxbury Press.
4. Fan, S., & Hyndman, R. J. (2012). Short-term load forecasting based on a semi-parametric additive model. Power Systems, IEEE Transactions on, 27, 134-141.
5. Hagan, M. T., & Behr, S. M. (1987). The time series approach to short term load forecasting. Power Systems, IEEE Transactions on, 2, 785-791.
6. Hippert, H. S., Pedreira, C. E., & Souza, R. C. (2001). Neural networks for short-term load forecasting: A review and evaluation. Power Systems, IEEE Transactions on, 16, 44-55.
7. Prautzsch, H., Boehm, W., & Paluszny, M. (2013). Bezier and B-spline techniques. Springer Science & Business Media.
8. Song, K. B., Baek, Y. S., Hong, D. H., & Jang, G. (2005). Short-term load forecasting for the holidays using fuzzy linear regression method. Power Systems, IEEE Transactions on, 20, 96-101.
9. Steadman, R. G. (1984). A universal scale of apparent temperature. Journal of Climate and Applied Meteorology, 23.
10. Taylor, J. W. (2003). Short-term electricity demand forecasting using double seasonal exponential smoothing. Journal of the Operational Research Society, 54, 799-805.
11. Tsay, R. S. (2005). Analysis of financial time series. John Wiley & Sons.