Grey-box Parameter Identification for Drive-train System of Large-scale Wind Turbine

Due to the resource limitation of fossil energies and the environmental requirements, wind energy has gained increasing attention in virtue of its richness in resources and cleanliness, and wind power generation technology has achieved rapid development. The modeling accuracy of wind turbines has a great influence on the dynamic analysis and control of wind turbines. Thus, this paper studies the dominant mechanism model of the drive-train system and analyzes its transfer function under the two-mass structure which is a non-self-balancing object. Then, two parameter identification methods via the recursive least squares (RLS) algorithm and the intelligent parameter identification algorithm with the particle swarm optimization (PSO) are compared. Furthermore, the grey-box state-space model of the drive-train system is constructed and the identification steps are systematically presented. In order to validate the effectiveness of the two methods, the identification processes are carried out based on the simulation data of a 5 MW wind turbine from FAST under different wind conditions. The identification results confirm the feasibilities of the proposed methods and prove that intelligent parameter identification is capable of acquiring better identification performances and less dependent on the selection and preprocessing of identified data.