In recent years，the rapid development of wind power has increasingly complicated the operation of power systems，with a higher risk of sub-synchronous oscillations （SSO）.The accurate and quick detection of SSO is highly important for effective countermeasures.However，the existing methods often exhibit poor noise adaptability and modal overlapping.To address these limitations，this paper proposes the parameterized multi-synchrosqueezing transform （PMSST），which combines weighted least squares，multi-synchrosqueezing transform，and polynomial chirp transform.PMSST first applies the multi-synchrosqueezing transform to achieve a high-energy-concentration time-frequency representation.The instantaneous frequency ridges of component signals are then extracted by ridge extraction algorithms，and Weighted Least Squares is employed to estimate the parameters of the transformation kernel.Finally，the time-frequency spectra are reconstructed to enhance the signal's energy representation，and rotation-invariant techniques are employed for parameter identification.According to simulation results，based on digital signals and doubly fed induction generator （DFIG） SSO simulations，PMSST effectively suppresses noise，accurately decomposes SSO signals，and yields reliable parameter identification.