A short-term wind power prediction approach based on ensemble empirical mode decomposition and improved long short-term memory

Wind power plays a significant role in the reduction of global carbon emissions. Traditional prediction approaches sometimes struggle to meet the actual needs due to the non-linearity, and volatility of wind power. Hence, an integrated prediction model is proposed, which combines long short-term memory (LSTM), ensemble empirical mode decomposition (EEMD), and particle swarm optimization (PSO). Firstly, the actual series was decomposed using EEMD into several subseries components with varying frequencies to mitigate the impact of the series' non-smoothness on prediction accuracy. Next, to mitigate the issue of subjectivity in the manual tuning of parameters in the traditional LSTM model, PSO is utilized to optimize hyperparameter values for LSTM. After constructing the PSO-LSTM model for each subseries, the ultimate prediction results are obtained by aggregating the predicted values from each subseries. The simulation results show that the proposed model achieves higher accuracy and stability of prediction than other comparative models.