To address the "curse of dimensionality" inherent in traditional sub-synchronous oscillation (SSO) analysis methods and to effectively mitigate SSO, we propose a damping ratio regression model alongside the design of a sub-synchronous damping controller (SSDC). Initially, the TLS-ESPRIT algorithm was employed to extract the principal modes of SSO. Subsequently, a damping ratio regression model was established using the principal component analysis (PCA)-random forest (RF) algorithm. The regression model was assessed to identify the dominant factors influencing SSO in wind farms. Thereafter, the SSDC was integrated into the static var generator (SVG) and optimized via a genetic algorithm (GA). Finally, a double-fed wind farm simulation model was implemented in PSCAD. The results demonstrate that the damping ratio regression model effectively identifies the dominant factors influencing SSO. The optimized sub-synchronous damping controller effectively resolves the issue of traditional SSDCs failing to provide positive damping under varying operating conditions and exhibits rapid suppression capabilities.

sub-synchronous oscillation (SSO),Damping ratio regression model,Modal identification,Sub-synchronous damping controller,Genetic algorithm