結合潛在資訊函數與倒傳遞類神經網路預測短期電力消耗量

Incorporating Latent Information Function in Back Propagation Neural Network for Predicting Short-Term Electricity Consumption

張孟仁(MENG-JEN CHANG)

指導教授 ： 張哲榮

中原大學/商學院/企業管理研究所/碩士(2014年)

https://doi.org/10.6840/cycu201400607

摘要

電力的總體需求量是國家發展的重要指標，也是電力系統規劃的重要方針，如何有效地對其預測，是政策擬定所必須面對的重要議題。在急速成長的開發中國家，由於具有不確定的國家經濟結構，這使得利用大量歷史觀察值進行預測的傳統方法不再適用。因此，能有效地利用少量樣本來獲得準確的預測結果，這對國家長期政策的擬定是一件極具意義的事。本研究於是試圖利用潛在資訊函數與倒傳遞神經網路來建立模型，以做為處理小樣本預測問題的方法；簡單地說，本研究先使用潛在資訊函數來分析資料行為，擷取資料的隱含資訊，再用資料的潛在資訊值來建立擴展的訓練集合，以學習倒傳遞神經網路的拓樸結構。經過亞太經濟合作會議能源資料庫之電力消耗量資料的實際測試，在三種常見的誤差指標中，本研究方法皆能有效地改善預測準確率；其中，平均絕對百分誤差僅為4.2795%，擁有良好的預測效果，顯見本研究提出的建模程序是一個處理小樣本資料的適當程序。

關鍵字

小樣本 ； 預測 ； 電力消耗量 ； 倒傳遞類神經網路

並列摘要

The overall electricity demand, regarded as a primary guideline for electricity system planning, is a widely used measurement for determining the degree of a nation’s development. Consequently, electricity consumption forecasts are critical for policy-making in rapidly developing countries. However, because the economic growth rates in these countries are typically high and unstable, obtaining accurate predictions by using long-term historical observations is difficult; thus forecasting with limited (short-term) data is highly effective and considerably appealing. This study applied the latent information function to analyze data features and extract hidden information for knowledge learning by using small datasets. The Asia-Pacific economic cooperation energy database was evaluated in the experiment, and the results show that the proposed method can significantly improve forecasting accuracy (the mean absolute percentage error in the datasets was 4.2795%). Thus, the proposed procedure is a feasible approach for small-dataset forecasting.

並列關鍵字

Electricity consumption ； Small data set

參考文獻

Liu, S. F., Lin, Y., Grey Information: theory and practical applications. Springer, London, Britain, 2006.

Li, D. C., Chang, C. J., Chen, C. C., Chen, W. C., Forecasting short-term electricity consumption using the adaptive grey-based approach-an Asian case. Omega-International Journal of Management Science, 40(6), 767-773, 2012b.

Abu-Mostafa, Y.S. Learning from hints in neural networks. Journal of Complexity, 6(2), 192-198, 1990.

Akay, D., Atak, M., Grey prediction with rolling mechanism for electricity demand forecasting of Turkey. Energy, 32(9), 1670-1675, 2007.

Amjady , N., Keynia, F., Short-term load forecasting of power systems by combination of wavelet transform and neuro-evolutionary algorithm. Energy, 34(1), 46-57, 2009.