ABSTRACT To address the issue of insufficient consideration of wind turbine physical characteristics and wind farm meteorological features in wind power forecasting, this paper proposes an ultra‐short‐term wind power prediction model based on digital twin technology. The model constructs a digital twin forecasting framework that integrates a digital‐physical model of the wind turbine and a parallel CTransformer‐BiGRU model to enhance prediction accuracy. The deep learning module captures spatiotemporal features in the data, while the digital‐physical model couples the forecasting process with the actual physical conditions of the wind farm, thereby improving prediction precision. Finally, the effectiveness of the proposed algorithm is validated through experimental tests on a real‐world dataset from a wind farm in Xinjiang, China.
Yang et al. (Wed,) studied this question.