采用神经网络与模糊曲线相结合的短期电力负荷预测方法

Hybrid short term load forecasting method using neural networks and fuzzy curve

文章提出了一种神经网络(ANN)与模糊曲线(Fuzzy Curve)有机结合的短期负荷预测方法,该方法采用ANN作为基本负荷预测,再用模糊曲线考虑影响负荷变化的因素(如天气的迅速变化、重大节假日等),对基本负荷预测做出修正,从而得到最终的负荷预测值。同时,针对传统BP学习算法的缺点,将BP算法和模拟退火算法的优点相结合以提高网络的学习性能。实例表明,该ANN—FC模型实用有效、精度高。

This paper proposes a hybrid short term load forecasting method which combines neural networks with fuzzy curve. The basic load forecasting is done by ANN. Factors, which have influences on load, are considered and the basic load forecasts are updated with fuzzy curve. As back-propagation learning algorithm has some drawbacks , BP&SA hybrid learning algorithm , which combines the property of BP with the property of SA algorithm , is presented to improve the learning property. Numerical tests showed the efficiency and accuracy of the ANN-FC method.