An integrated framework for adaptive time-frequency characterization of offshore wind turbine dynamics during extreme events

• Integrated AMD/MVMD, kurtosis filter, EPSDT, and ASLT for OWT dynamic analysis; • Enables high-resolution frequency tracking under noise and typhoon loads. • Validated by simulations and field data under extreme wind conditions. • Reveals amplitude–frequency stabilization as a nonlinear behavior indicator. The global expansion of offshore wind energy under decarbonization initiatives has heightened the need for precise monitoring of turbine dynamics. Extreme environmental loads induce rapid, nonlinear variations in the natural frequencies of offshore wind turbines, which conventional time–frequency methods often fail to capture due to fixed resolution, noise sensitivity, and limited real-time capability. This study introduces an adaptive framework integrating modal decomposition, kurtosis-based noise suppression, and the Adaptive Superlet Transform to enable high-resolution, noise-resilient tracking of time-varying modal frequencies under extreme conditions. The framework employs modal decomposition to isolate modal responses, a kurtosis criterion to suppress blade-passing harmonics, Enhanced Power Spectral Density Transmissibility to mitigate colored noise, and the Adaptive Superlet Transform for accurate instantaneous frequency estimation. Numerical simulations on a three-degree-of-freedom model yield frequency identification errors below 5% under both stationary and non-stationary conditions. Field tests on an offshore turbine during Typhoon Talim further validate the method’s capability to capture evolving modal behavior. Application to a real-typhoon event reveal amplitude–frequency stabilization of the fundamental mode under extreme loads. This stabilization provides a novel foundation for developing quantitative indicators to support real-time structural health monitoring and reliability assessment of offshore wind turbines.