融合VMD与FECAM的日前电价预测研究OA

In response to the challenges of frequent fluctuations,strong nonlinearity and drastic extreme value changes in the electricity spot market,this paper proposes an electricity price forecasting model that integrates variational mode decomposition(VMD)and the frequency enhanced channel attention mechanism(FECAM).It firstly decomposes the original electricity price series into multiple intrinsic mode functions with distinct frequency components by using VMD,effectively reducing data non-stationarity.Then,it uses convolutional neural networks(CNN)to extract key features and combines bidirectional long short-term memory network(BiLSTM)to enhance long timing process ability of the model.Finally,by introducing the FECAM,the model's adaptability to critical features is strengthened.The experimental results demonstrate that the proposed model outperforms traditional regression models and other deep learning approaches on datasets from the Australian electricity market,GEFCom2014,and the U.S.PJM market.The model exhibits superior prediction accuracy and demonstrates its applicability in complex electricity market environments.