Hybrid time–frequency method for vibration fatigue damage analysis under non-stationary non-gaussian random excitation

Abstract

Accurate assessment of vibration fatigue life under non-stationary non-Gaussian excitations remains a critical challenge in mechanical system reliability. Traditional vibration fatigue frequency domain methods relying on stationary power spectral density (PSD) inherently overlook temporal variations and non-Gaussianity in stress responses, resulting in fatigue damage prediction errors that escalate exponentially with the fatigue exponent b. To address this limitation, this study proposes a hybrid time–frequency methodology that incorporates the amplitude-frequency response characteristics of the structure and reveals the critical role of excitation signal phase in influencing vibration fatigue life. The method not only resolves the inherent shortcomings of conventional spectral approaches but also enables inverse synthesis of excitation signals with controllable response characteristics. Numerical simulations and experimental validations demonstrate that the proposed approach significantly reduces in fatigue life predictions. Furthermore, the method serves a dual purpose: it advances vibration fatigue analysis for systems operating in complex service environments and introduces a novel approach for designing accelerated fatigue tests through tailored excitation signals. This work attempts to narrow the gap between non-stationary non-Gaussian vibration theory and practical engineering applications, providing a useful tool for reliability-driven mechanical design.