Dynamic multi-turbines spatiotemporal correlation model enabled digital twin technology for real-time wind speed prediction

Abstract

The spatial and temporal correlation prediction model, because of its high accuracy, has become the mainstream direction in wind speed prediction field. However, the time-varying and unpredictability of wind speeds are still complex problems that negatively affect prediction result. Thus, this work aims to solve such challenges by developing a predictable multi-turbines spatiotemporal correlations framework (PMTSTCF) assisted digital twin and Internet of Things technologies. Firstly, a synthetic architecture integrating dynamic screening for correlated turbines and synchronized verification for prediction results is constructed. Then, the multi-turbines correlations model, which employs the propagation time delay and spatial similarity of wind energy on its motion paths to formulate real-time wind speed prediction task, is proposed. Meanwhile, a multi-variables verification and feedback mechanism is designed to synchronously track the spatiotemporal correlations among turbines and optimize the combinations of multiple correlated reference turbines. Finally, the predicted wind speeds are obtained by dynamical fusion of multiple single seed correlation prediction results that are predicted by leveraging the spatiotemporal dependencies. In case study, two algorithms, including support vector regression and Kalman filter, are employed to validate the effectiveness of the proposed PMTSTCF. The results demonstrate that PMTSTCF can furtherly improve accuracy and robustness of prediction model.