摘要
利用灰色理论中累加生成方法能够削弱负荷中随机成分的特点,以及人工神经网络可以逼近任意函数的能力,对具有任意变化规律的数据序列进行拟合和预测.实验结果表明,基于灰色理论和神经网络的最优组合模型的平均相对误差为1.307%,比BP神经网络预测和灰色理论模型预测的精度更高,具有明显优势.
The accumulated generating method of gray theory can weaken the random ingredients of the load, and artificial neural networks can be adjacent to any function, a sequence which changes arbitrarily is fitted and forecasted. The experimental results show that the average relative error based on gray theory and neural network model for the optimal combination is 1. 307%, and this method has obvious advantages in forecast precision over BP neural network forecast and gray theory model forecast.
出处
《上海电力学院学报》
CAS
2013年第6期527-531,共5页
Journal of Shanghai University of Electric Power
基金
上海市教育委员会创新基金(11YZ192)
