用于短期电力负荷预报的人工神经网络方法

The Artificial Neural Network Approach for Short-term Load Forecasting

提出了一种新的短期电力负荷预报方法。该法首先将小时电力负荷分解成增长趋势、日模式、周模式、气候敏感及随机变动等分量，然后应用人工神经网络等方法对各分量分别预报，最后由各负荷分量相迭加得到小时负荷预测值。并以我国某省实际电力系统负荷预报为例，说明了所提方法的有效性。

This paper has presented a new approach for short-term load forecasting.Thehourly electric load is decomposed into daily pattern component、weekly pattern component、growthtrend、weather sensitive component and stoctiastic variation component. The artificial newralnetworks are employed to predic the trend and weather sensitive components. The proposedapproach has been tested on a proctical power system and the results are satisfactory.