Efficient Fatigue Reliability Life Analysis Method Based on Sequential Adaptive Surrogate Model

To efficiently estimate the fatigue reliability life, this paper proposes an efficient sequential adaptive semi-surrogate model-based approach. First, the analysis of fatigue reliability life involving complex inverse fatigue life reliability analysis is equivalently transformed into a series of forward fatigue life reliability analyses under different fatigue life thresholds using the dichotomy method. Second, an adaptive kriging surrogate model is established that solely substitutes the structural analysis model generally constructed by implicit finite element analysis named as the semi-surrogate method, with an approximate analytical solution for the error transfer function from the kriging-model-based finite element analysis model to the corresponding limit state function, thereby achieving rapid updating of the kriging model. Finally, a sequential updating criterion for kriging model of the structural analysis model under varying fatigue life thresholds is developed, in which training samples are shared among kriging models in the series fatigue life reliability analyses. The efficiency and accuracy of the proposed method are verified through fatigue reliability life analyses of the aeroengine turbine disk and turbine shaft.