Wind power forecasts through hybrid model based on VMD and Wavelet decomposition

Wind power prediction is challenging due to its dependence on diverse weather factors like wind speed, humidity, and atmospheric pressure. This study proposes a novel hybrid approach combining Variational Mode Decomposition (VMD) and Wavelet Transform for signal preprocessing, followed by a Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) model for prediction. Performance evaluation using multiple metrics demonstrates that the VMD-CNN-LSTM model outperforms alternatives, achieving RMSE values of 0.58165 and 0.39742 on two benchmark datasets. The proposed hybrid architecture was effectively able to predict wind power with high accuracy and reliability.