With the increasing penetration of renewable energy into power grids, wideband-oscillation issues have become increasingly prominent. In practical wind farms, the key control parameters of aging or imported wind turbine generators (WTGs) are often difficult to obtain directly, leading to challenges in accurate system modelling. Since oscillation characteristics are closely related to the key control parameters of wind power converters, identifying these parameters from limited on-site oscillation data is a feasible approach. To address this challenge, this paper proposes a data-driven method for identifying key control parameters of doubly-fed induction generator (DFIG)-based WTGs using sub-synchronous oscillation data. The method requires only limited field-recorded oscillation data to achieve parameter identification, eliminating the need for control system modifications or external excitation injections. Taking a DFIG-based wind farm integrated with a series-compensated grid as a case study, the Hilbert-Huang Transform (HHT) is employed to extract dominant oscillation mode features (frequency and damping ratio). Phase margin sensitivity analysis is then applied to determine key control parameters. A neural network model is constructed with output active power and oscillation characteristics as inputs and key control parameters as outputs. The identification accuracy is validated using measured oscillation data. The proposed method offers a novel solution for “grey-box” WTG control parameter identification, demonstrating significant practical value.
Wang et al. (Sun,) studied this question.