Non‐linear autoregressive moving average with exogenous input model‐based adaptive control of a wind energy conversion system

Abstract

Wind energy conversion system (WECS) is a stochastic system, since wind speed varies intermittently. Therefore a non-linear autoregressive moving average with exogenous input (NARMAX) model is developed to represent the dynamics of WECS which is used for real-time implementation. In a doubly-fed induction generator (DFIG) WECS, speed and power are the outputs for regulation which are achieved by controlling the torque and pitch angle, repetitively. NARMAX model identifies the structure and significant terms of speed and power of DFIG WECS and its parameters are estimated employing an on-line adaptive recursive least squares algorithm. For optimisation of torque and pitch angle, performance index (PI) is defined in non-linear adaptive model predictive controller (NAMPC) to achieve the control objective, i.e. torque and pitch angle. The weights in PI are updated until the optimised values in control inputs (torque and pitch control) are achieved. Boundedness of the WECS is defined by considering the constraints on the outputs and control inputs. Extensive simulations are carried out with NARMAX structure with NAMPC on DFIG WECS using MATLAB/SIMULINK and the performance is compared with conventional proportional–integral controller and model predictive control. From the obtained results, it is observed that the NARMAX model with NAMPC has minimum deviations from the operating point in power, speed, torque and pitch angle compared to other controllers.