Adaptive control of a wind turbine based on neural networks

In this paper, we investigate on Indirect Model Reference Adaptive Neuro Control (IMRANC), for output electrical power tracking of a nonlinear non-affine Horizontal Axis Wind Turbine (HAWT). The nonlinear system is first identified by the Nonlinear Autoregressive network with Exogenous inputs (NARX) model that is a recurrent dynamic network. Afterward an IMRANC is designed based on NARX identified model to reach the close loop system to desired reference model. The MLP networks are applied for both of model and controller subsystems and are then trained by the Marquardt-Levenberg Back-Propagation (LMBP) algorithm while the Tapped Delay Lines (TDL) components are considered over input and feedback paths. Simulation results are final presented to validate the effectiveness of the proposed method like robustness and good load disturbance attenuation and accurate tracking, even in the presence of parameter variations due to changing of hydraulic pressure in hydraulic pitch system and also disturbances on the system.