摘要
针对目前电力负荷预测算法精度不高的现状,提出使用Ada-BP神经网络改进算法作为负荷预测的新方法。通过对同一个训练集训练不同的弱学习器,然后将这些弱学习器集合起来,构成一个强学习器,从而提高算法的泛化能力以及预测精度。将此算法应用于某区域实际电网,结果表明该改进算法满足当前区域电网对负荷预测精度的要求,比常用算法表现出更好的泛化能力,具有一定的实际应用价值。
In view of the present electric power load prediction algorithm with low precision, this paper puts forward a Ada-BP neural network improved algorithm for load forecasting, trains different weak learning machines of the same training set, and then assembles the weak learning machine together to constitute a strong learning machine so as to improve the generalization ability of the algorithm and the prediction precision. This algorithm is applied to a practical grid, the results show that the improved algorithm meets accuracy requirements for the load forecasting and show better generalization ability comparing with common algorithm.
出处
《陕西电力》
2012年第12期21-24,共4页
Shanxi Electric Power
基金
中国博士后科学基金资助项目(20100451063)
关键词
负荷预测
神经网络
电力负荷
泛化能力
load forecasting
neural network
power load
generalization ability
