The Research of Wind Farm Power Forecasting Based on VMD-IPSO-CNN-LSTM Hybrid Approach

To enhance the accuracy and stability of wind-farm power forecasting, this study proposes a hybrid framework integrating Variational Mode Decomposition (VMD), an improved Particle Swarm Optimization (PSO) algorithm, Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM) networks. First, we analyze the relationship between wind-power time series and their key influencing factors to guide the selection of VMD parameters, decomposing the original data into multiple intrinsic mode components. Second, we employ a multiscale feature-extraction module based on CNN to capture spatial patterns, and a LSTM-based module to model temporal dependencies. The improved PSO is then applied to fine-tune critical hyperparameters—such as the number of CNN convolutional layers and LSTM hidden units—thereby improving both predictive accuracy and robustness. Finally, comparative experiments under varied meteorological scenarios and abrupt wind-speed changes demonstrate that the proposed method consistently outperforms benchmark models in terms of accuracy and stability. These results confirm its capability to deliver efficient and reliable support for wind-farm power prediction.