基于CEEMDAN分解的风电功率预测方法

Wind Power Prediction Method Based on CEEMDAN Decomposition

为了解决由于风电中出现的随机性和波动性对风电功率预测造成的预测精度不高的问题,提出了一种基于CEEMDAN的风电功率预测方法。首先,使用CEEMDAN算法对风电功率进行信号分解,然后,在LSTM算法的基础上,使用Encoder-Decoder框架,结合Attention机制,设计了ELDGAWP风电功率预测模型,有效解决了LSTM在处理非常长的序列数据时,模型可能会遇到梯度消失的问题,最后将各模态分量预测结果累加得到最终预测结果。经与现有模型进行对比,本文所提C-ELDGAWP预测方法的预测精度最高。

To address the issue of low prediction accuracy in wind power forecasting caused by the randomness and volatility in wind energy,a wind power prediction method based on CEEMDAN is proposed.Firstly,the CEEMDAN algorithm is used to decompose the wind power signal.Then,utilizing the LSTM algorithm as the foundation,an ELDGAWP wind power prediction model is designed by incorporating the Encoder-Decoder framework and Attention mechanism.This effectively solves the problem of gradient vanishing that LSTM models may encounter when dealing with very long sequence data.Finally,the predicted results of each mode component are accumulated to obtain the final prediction result.Compared with existing models,the proposed C-ELDGAWP prediction method achieves the highest prediction accuracy.