Fatigue crack damage tolerance life prediction based on SCN-IHDMR method

The damage tolerance life calculation can accurately assess the remaining life of fatigue cracks. However, when the spatial parameters vary randomly, it is difficult to evaluate and predict the remaining fatigue life efficiently. Therefore, the stochastic configuration network(SCN) method and the improved high -dimensional model representation(IHDMR) technique are used to construct a surrogate model, which can accurately predict the remaining life of fatigue cracks coming from different spatial parameter variations. A sample library of surrogate models for fatigue crack damage tolerance life(FCDTL) prediction was determined by mutual validation of experiments and finite element analysis(FEA). The surrogate model considers factors such as stress level, crack morphology, material properties and surface treatment, and equivalently transforms the above factors into characteristic parameters. The results reveal the surrogate model basing on SCN algorithm and dimensionality reduction technique has accurate prediction effect and high computational efficiency for FCDTL.