基于经验模式分解的神经网络组合风速预测研究

Research on Neural Network Combined Wind Speed Prediction Based on Empirical Mode Decomposition

为了提高风电并入电网的安全性,需要对风功率进行提前预测。风速预测是风功率预测的关键,而风速的不稳定性是预测的难点。为了降低风速的不稳定性,提高预测精度,提出经验模式分解法将风速分解并重组成2组不同的序列,对高频分量采用神经网络组合预测,剩余分量采取BP神经网络预测,并对两分量预测结果等权相加得预测结果。针对不同的样本进行建模预测,验证了该方法的适用性。并比较了GRNN、BP、LS-SVM 3种方法不同组合方式的预测精度,证明了在该组合方法中3种方法优势互补。

In order to improve the safety of the wind power integrating into the power grid,it is necessary to predict the wind power ahead of time. The wind speed prediction is the key to wind power prediction,and the instability of wind speed becomes the conundrum for the prediction. In order to reduce the instability of wind speed and improve prediction accuracy,an empirical mode decomposition method is proposed to decompose the wind speed into two groups of different sequences,and the high frequency components are predicted by neural network combination while the residual components are predicted by BP neural network. The prediction results are obtained by adding the two components calculated by the above two networks respectively. The applicability of the method is verified by modelling different samples.And three different methods,GRNN,BP,and LS-SVM,are combined in different ways to conduct the prediction,and the accuracies are compared. The results show that they have their own advantages with different combinations,which confirms that these three methods could complement each other.