Wind energy conversion system model identification and validation

Abstract

Wind energy conversion system (WECS) is complex because of wind speed varies in time and space. Model identification is required to represent its dynamics for real-time implementation. In this paper a doubly-fed induction generator (DFIG) WECS is used. Different model structures are generated and simulated using MATLAB/SIMULINK. The models are generated using both nonlinear and linear system identification tool boxes. Linear system identification toolbox generates both model structure and model parameters; whereas the nonlinear system identification tool generates only the system model structures. From linear models, the BJ33221 model has better performance with best fit of 74.78%, final prediction error (FPE) value of 0.0445 and mean square error (MSE) is 0.04265. ARX211 model structure provides best fit of 74.39%, FPE of 0.0453, and MSE is 0.04465. This study shows as model order increases, the best fit value too, but the system become more complex. The nonlinear models have better performance than the linear models. The nlarx121 model structure provides the best fit of 96.43% and MSE of 0.0322, with other technique for its model parameters estimation. The output residuals are within the confident range (0.2 to −0.2), indicating the model structure was validated.