采用小波分析和神经网络的短期风速组合预测

Short-Term Wind Speed Combined Forecasting Using Wavelet Analysis and Neural Network

为了提高风速的波动性与随机性预测精度,提出小波分析和神经网络组合的风速预测模型.该方法利用小波分解将风速分解为一列频率不相同的分量,并利用二插值进行重构;根据各个分量的频率特征,选择合适的模型分别进行预测;高频分量采用组合神经网络预测,低频分量采用合适的单一模型直接进行预测;将各预测值叠加得到最终预测值.算例分析表明:相较于单一预测模型,所提方法的预测精度得到大幅提升,更加贴近实际风速曲线,预测结果更具可靠性.

In order to improve the forecasting accuracy and reduce the impact of randomness and volatility of the wind speed,a new forecasting model based on wavelet analysis and neural network is presented.By means of the wavelet analysis technique,the original wind speed series are decomposed into a series of components of different frequencies components,and reconstructed by twice-interpolation.According to the frequency characteristics of each component,the suitable model is selected to forecast separately.The high frequency component is forecasted by combined neural network,and the low frequency component is forecasted by a suitable single model.The final forecasting value is obtained by superimposing each forecasting values.The simulation results show that compared with the single prediction model,the forecasting accuracy of the proposed method is greatly improved,and it is closer to the actual wind speed curve,and it is more reliable.