基于模态分解与多尺度混合的短期风电功率预测OA

As the proportion of wind power in power systems continues to increase,short-term wind power prediction is crucial for ensuring the stable operation of power grids.For this reason,a wind power prediction model based on the complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)and the improved multiscale hybrid recurrent neural network(TimeMixer-GRU)is proposed.Firstly,the CEEMDAN technique is applied to decompose the original wind power sequence into subsequences with different frequency characteristics,so as to reveal its inherent and complex fluctuation patterns.Then,for each subsequence,a decomposable multiscale hybrid network is adopted to conduct pooling down-sampling at different levels to obtain sequences at different scales.Seasonality and trends are extracted according to the decomposition of sequences at different scales to obtain multiscale temporal features.Finally,different scale sequences of each subsequence are predicted by the recurrent neural network,and the predicted values of each subsequence are recombined to output the prediction results.Through the experimental analysis of the real wind power data,the results show that this modal decomposition and multiscale hybrid method can effectively extract the key information of the wind power sequence,improve the accuracy and stability of short-term wind power prediction,and provide powerful technical support and a decision-making basis for the efficient utilization of wind power resources and the reliable dispatching of power grids.