Very Short-Term Forecasting of Wind Power Based on Transformer

Accurate wind power forecasting is crucial for the stability of modern power systems and fostering wind power utilization. However, very short-term forecasting faces challenges due to its limited input duration, and the utilization of long sequences is rarely employed in this context. The reason behind this limitation lies in the fact that traditional forecasting models often encounter the issues of gradient disappearance or gradient explosion when handling long sequences. Therefore, this paper presents a novel very short-term wind power forecasting model based on Transformer (TF), aiming to explore the feasibility of utilizing long sequences for very short-term forecasting. The proposed model is evaluated through case using real-world engineering data. The obtained numerical results demonstrate that TF is capable of effectively processing long sequences, providing valuable insights for the advancement of future forecasting models.