Fatigue life prediction method based on polynomial chaotic expansion and Weibull distribution

This study proposes a fatigue life prediction method combining small-sample data expansion with the Weibull distribution function, incorporating the first order reliability factor (FOSM) to improve accuracy. Using Generalized Polynomial Chaos Expansion (GPC) and Latin Hypercube Sampling (LHS), small-sample fatigue data is expanded, followed by enhancing the two-parameter Weibull model with FOSM. Results show the generalized polynomial chaotic expansion method and Latin hypercube sampling are used to obtain the probability density curve when the stress level is 350 MPa, and the original data are all on this probability density curve, indicating that the expansion method is more credible. High prediction precision within a 1.5 × error range, with logarithmic safety life linearly related to stress level and decreasing with higher failure probability.