考虑尾流延迟特性的海上风电场LPV模型预测控制

The pronounced wake effect in large-scale offshore wind farm necessitates careful consideration of its delay characteristics, which are crucial yet often overlooked in control. Addressing the coupling between the dynamic evolution of wake effects and the parameter changes of the WT control model, this paper introduces a Linear Parameter-Varying (LPV) model predictive control method that considers wake delay characteristics. Through the development of a quasi-steady state wake model, the wake delay characteristic is incorporated within an LPV model for the wind farm. A two-stage dimensionality reduction method is proposed to simplify the calculation, and a model predictive control (MPC) method is combined to optimize the fatigue damage balance and power generation enhancement in the wind farm coordinately. Simulation results from a wind farm consisting of 16 wind turbines validate the efficacy of the quasi-steady state wake model in depicting the spatial distribution of wake delays. Furthermore, in dynamically varying wind speed and directions scenarios, the proposed control method can effectively capture the wind speed delay and fluctuation characteristics between different wind turbines, leading to heightening power output and diminishing fatigue stresses. Notably, in comparison to the static model control, the adaptive parameter tuning mechanism inherent in the proposed method effectively curtails control overshoot, enhancing overall system performance.