摘要
由于基于反向传播(back propagation,BP)的神经网络模型自身固有的缺点,其电力负荷预测结果不理想,而径向基函数(radial basis function,RBF)神经网络模型具有全局逼近的性质,不存在局部最小问题,为此,针对中长期电力负荷预测,给出了RBF的预测原理,推导权值的更新方式,并和BP方法结果进行对比分析,结果证明基于RBF神经网络模型的方法收敛速度快、预报精度高、误差小。
The power load forecasting results of neural network model based on back propagation (BP) are not satisfactory owing to the inherent shortcomings of the model. The radial basis function (RBF) neural network model, with the nature of global approximation, has no local minimum problem. Aiming at medium- and long-term power load forecasting, the forecasting principle of RBF is presented and the updating expressions of weights are derived. By means of calculation example, the RBF method and the BP method are compared, and comparison results show that the RBF method is with fast convergence, high precision and small error.
出处
《广东电力》
2010年第5期1-3,11,共4页
Guangdong Electric Power
基金
国家自然科学基金资助项目(50677014)
博士点专项基金资助项目(20060532002)
湖南省自然科学基金资助项目(06JJ2024
03GKY3115
04FJ2003
05GK2005)
关键词
反向传播神经网络模型
径向基函数神经网络模型
负荷预测
back propagation (BP) neural network model
radial basis function (RBF) neural network model
load forecasting