Energy capture efficiency enhancement for PMVG based-wind turbine systems through yaw control using wind direction prediction

Accurate wind direction prediction is fundamental for the efficient operation of wind turbines and is also important for optimizing the performance and efficiency of the wind turbine system (WTS). In this study, we present a wind time-series-based prediction technique using a deep neural network (NN) approach to predict the wind direction and also aim to do the maximum power extraction (MPE) of a permanent magnet vernier generator (PMVG)-based WTS using the proposed model predictive control (MPC)-based yaw control method to improve its energy capture efficiency. To do this, an echo state network (ESN) approach is designed with a non-linear function and extended Kalman filter (EKF) to handle the non-linearities and improve prediction accuracy by eliminating noisy measurements, thus predicting the wind direction at an effective rate. Next, the performance of the proposed direction prediction model is compared with other prediction methods. The predicted wind direction is utilized in a finite control set model predictive control (FCS-MPC)-based yaw control strategy, enabling optimal turbine alignment and maximizing energy capture efficiency. Finally, superiority and robust performance of the proposed controller are evaluated and compared to existing control methods such as proportional-integral (PI), proportional-integral-derivative (PID) and baseline MPC using simulation of 4.8 MW PMVG-based benchmark WTS.