基于小波分解的神经网络组合风速预测

Wind Speed Prediction Through Neural Network Combination Based on Wavelet Decomposition

短期风速预测对于风电机组一次调频有着重要意义,而风速的随机性和波动性会直接影响到风速预测的精度。针对风速的上述特点提出了一种基于小波分解的神经网络组合风速预测方法。首先通过小波分解将不稳定的风速信号进行分解,从而得到不同频率的分量并进行重构;然后对高频分量分别采用Elman、BP神经网络预测并选取合适的权重比进行加权平均,低频分量采用Elman神经网络进行预测;最后将每一个预测的值进行叠加获得最终的预测值。算例分析表明:基于小波分解的神经网络组合预测方法的预测精度有很大提高,预测结果更具有可靠性。

Short-term wind speed prediction is of great significance for the primary frequency modulation of wind turbines,and the randomness and fluctuation of wind speed will directly affect the accuracy of wind speed prediction.In view of the above mentioned characteristics of wind speed,a wind speed prediction method through neural network combination based on wavelet decomposition was presented in this paper.Firstly,unstable wind speed signal was decomposed through wavelet decomposition so as to obtain different frequency components and complete reconstruction.Then,high-frequency components were predicted by use of Elman and BP neural networks respectively,and appropriate weight ratios were selected to obtain weighted average.Low-frequency components were predicted by use of Elman neural network.Finally,each predicted value was superimposed so as to obtain the final predicted value.The analysis of examples indicated that the accuracy of the neural network combination prediction method based on wavelet decomposition was greatly improved and prediction results became more reliable.