基于经验模态分解与RBF 神经网络的短期风功率预测

Short-term Wind Power Prediction Based on Empirical Mode Decomposition and RBF Neural Network

风功率预测的准确性对优化电力系统调度、促进新能源消纳与增强电力系统稳定性有重要意义。针对风功率预测准确度问题,提出了一种基于经验模态分解与径向基神经网络的短期风功率预测方法。首先,采用小波变换对风功率历史数据进行去噪处理,利用经验模态分解将去噪后的历史数据分解为多个模态分量序列。然后,在考虑多种气象因素的条件下分别构建径向基神经网络对各分量序列进行分项预测。最后,叠加各网络输出得到预测结果。基于张家口某风电场的风功率与气象数据,以均方根误差和平均绝对百分误差作为评价指标对本算法进行测试。结果表明,分解后的风功率分量序列具有较强的规律性,预测精确度高于其他4种传统预测算法,证明了本算法的有效性。

The accuracy of wind power prediction plays an important role in optimizing the power system dispatching,improving the renewable energy accommodation and enhancing the power system’s stability.In view of the accuracy of wind power prediction,a short-term wind power prediction method is proposed in this paper,which is based on empiri⁃cal mode decomposition(EMD)and radial basis function(RBF)neural network.First,wavelet transform is utilized to denoise the historical wind power data,and the de-noised historical data are further decomposed into several modal component sequences using EMD.Then,RBF neural networks are constructed considering various meteorological fac⁃tors,thus forecasting each component sequence separately.Finally,the prediction result is obtained by superimposing the output from each network.Based on the wind power and meteorological data of one wind farm located in Zhangjiak⁃ou,root-mean-square error and mean absolute percentage error are used as evaluation indices to test the proposed algo⁃rithm.Results show that the decomposed wind power component sequence has strong regularity,and the prediction ac⁃curacy is higher than those of the other four traditional prediction algorithms,which proves the effectiveness of the pro⁃posed algorithm.