Adaptive real-time wind farm control based on ultra-short-term wind prediction and wake dynamics

The highly random and time-varying nature of realistic wind conditions underscores the need for accurate short-term wind speed prediction and dynamic wake characterization to achieve effective real-time wake control in wind farms. This paper presents a novel real-time cooperative control system that integrates data-driven short-term wind speed predictions into a dynamic wind farm control framework. The prediction accuracy of the fractal interpolation extrapolation method is evaluated using measured wind data, and a six-turbine wind farm is selected to test the performance of the proposed system against traditional methods. Additionally, a feasibility analysis is conducted with consideration of various control intervals. Simulation results demonstrate that the improved fractal interpolation prediction method excels in short-term wind speed prediction, with a prediction error of only 5.65%. Dynamic wind farm control based on fractal interpolation prediction significantly outperforms control based on commonly used look-ahead data in terms of power benefits. However, this advantage is diminished in steady-state wind farm control. The proposed control system achieves a power increase of 3.233%, approaching the theoretical optimum, and substantially surpasses the 0.414% increase achieved by traditional steady control based on look-ahead data. The performance of the proposed control system remains robust until the error from fractal interpolation prediction, which accumulates with the control interval, reaches a certain threshold.