Integrated Driving Method to Optimize Operation of Wind Turbines for Weakening the Wake Effect

The wake effect in wind farms (WFs) leads to wind energy loss and increases the fatigue load of wind turbines (WTs). In this paper, an integrated driving method (IDM) to optimize operation of wind turbines (WTs) for weakening the wake effect. Considering the wake effect with delayed time, a model predicted control (MPC)-based WT control method is proposed to optimize the dual-parameter, i.e. pitch angle and generator torque, for mitigating fatigue load and increasing power generation. To pursue the better optimization effect, the data-driven method (DDM) is proposed to learn the accurate and complete dataset for eliminating linearization errors in the nonlinear physical of wind turbines and engineering wake models. DDM is designed to capture the nonlinear aerodynamic interactions based on the operational status of WTs in wake-affected WFs. It utilizes a convolutional neural network (CNN)-based encoder and a graph neural network (GNN)-based decoder. Subsequently, an artificial neural network (ANN)-based fusion method is proposed to select the better optimization values of these two methods for improving the prediction reliability and anti-measurement noise interference ability. Finally, the efficacy of the proposed scheme is substantiated through case studies involving a wake-affected WF.