Abstract: Non-Gaussian random stresses exist in engineering stuctures widely, but the former spectral fatigue life estimate methods cannot be efficiently applied to this kind of stresses due to basing on Gaussian assumption. Based on spectral fatigue life estimate theory and Hermite polynomial extension of non-Gaussian Possibility density function and three dimensions high order statistic, a new method is theoretically deduced to estimate fatigue life under non-Gaussian random stresses, and is demonstrated with numerical examples under Gaussian stress, sub-Gaussian stress and super-Gaussian random stresses. The method is applicable to fatigue estimate under stresses whether it is Gaussian or non-Gaussian, and whether its spectral bandwidth is narrow or wide.

Key words: Non-Gaussian Random stress, Power spectral density, Life estimate, Spectral fatigue