Model identification and analysis of small-power wind turbines

In this paper, we present complete results for model identification methods and analysis of a small-power wind turbine in the prospect of designing efficient controllers for obtaining maximum electrical power output and devising the fault detection and diagnosis schemes. The system has been identified using three different model structures: ARX, ARMAX and state-space models. The techniques used for their estimation are least-squares, prediction-error and subspace-based N4SID methods, respectively. Identification and validations are performed on actual measurements of a wind turbine installed at West Michigan University (WMU). It is concluded that the identified ARX model gives the best results in terms of minimum value of Akaike's information criterion (AIC) and maximum percentage of fitness when validation tests are performed.