Fatigue reliability analysis for structures with hybrid uncertainties combining quadratic response surface and polynomial chaos expansion

This study attempted to propose a novel efficient and accurate uncertainty analysis method for structural fatigue life prediction with hybrid uncertainties. First, PCE was adopted to deal with independent normal random variables, where polynomial coefficients were expressed as quadratic functions of interval variables. In order to determine unknown coefficients, a SCC collocation technique was applied combining with LHS. Interval bounds of the corresponding mean value and standard variance were then deduced, after which a fatigue reliability analysis method was proposed. Finally, three numerical examples were implemented to validate the feasibility and efficiency of the proposed method.