Newton Iteration Method-Based Online Inductance Identification for MRAS Position Observer of Variable-Speed Pumped-Storage

This paper presents an online dual inductance identification method for the model reference adaptive system (MRAS) position observer of a megawatt doubly-fed induction generator (DFIG) in a variable-speed pumped-storage (VSPS) unit. The adaptive regulator is designed using a novel method, i.e., the Newton iteration method, which requires very few calculations and imposes a low computational burden on microprocessors in mutual inductance identification during each control step. The proposed Newton iteration method exhibits second-order convergence ability and is guaranteed to converge to a stable state for any positive initial input value. In addition, the analytical parameter sensitivity formula is derived, which quantitatively expresses the relationship between the mismatched inductance and the steady-state error of the MRAS position observer. Moreover, the observation accuracy of the MRAS position observer is improved by applying the proposed method. Finally, the performances of the suggested method are thoroughly investigated and validated by simulation results and experimental works.