A Characterization Method for Fatigue Damage under Broadband Non-Stationary Non-Gaussian Random Vibrations

Broadband non-stationary and non-Gaussian (NSNG) random vibrations widely exist in the service environments of complex mechanical systems. The rainflow counting method is considered the most accurate approach for evaluating fatigue damage under such conditions; however, it becomes computationally expensive when handling long-duration or high-frequency vibration data. To address this limitation, this study investigates the relationship between the statistical characteristics of broadband NSNG random signals and their fatigue damage obtained by rainflow counting. The results show that when the material fatigue exponent b is small, a positive correlation exists between the b-th order absolute moment of the vibration signal and its rainflow-counted fatigue damage. Based on this finding, a rapid fatigue evaluation method is proposed that bypasses rainflow counting by directly using the b-th order moment to characterize fatigue damage under broadband NSNG random vibrations. This approach provides a computationally efficient alternative for fatigue assessment in stochastic vibration environments.