Wind speed prediction based on improved Informer

With the increasing depletion of global energy, wind energy, a green and clean energy source, is receiving increasing attention. To improve the accuracy of wind speed prediction, this paper proposes a hybrid prediction model based on VMD-FECAM-TCN Informer. In response to the non-stationarity and multi-scale characteristics of wind speed sequences, Variational Mode Decomposition (VMD) is introduced to decompose the signal, resulting in modal components with good time-frequency localization properties, effectively separating different frequency components. Next, the frequency-enhanced channel attention mechanism (FECAM) weighs each modality’s importance, enhancing the model’s ability to capture crit features. Then, the Time Convolutional Network (TCN) is used to solve the problem of long-term dependencies over long distances. Finally, the features are input into the Informer model for long-term prediction, fully leveraging its advantages in capturing long-term dependencies. This article conducted sufficient ablation experiments on the 2018 Scada Data of a Wind Turbine in Turkey, and the experimental results validated the effectiveness of the proposed method.