A hybrid prognosis model for predicting fatigue crack propagation under biaxial in-phase and out-of-phase loading

Abstract

A hybrid prognosis model has been developed to predict the crack propagation in aluminum alloys subject to biaxial in-phase and out-of-phase fatigue loading conditions. The novel methodology combines physics-based modeling with machine learning techniques to predict crack growth in aluminum alloys. Understanding the failure mechanisms under these complex loading conditions is critical to developing reliable prognostic models. Therefore, extensive fatigue tests were conducted to study the failure modes of carefully designed cruciform specimens. Energy release rate was used as the physics-based parameter and Gaussian process was used to model the complex nonlinear relationships in the prognosis framework. The methodology was used to predict crack propagation in Al7075-T651 under a range of loading conditions. The predictions from the prognosis model were validated using the data obtained from the biaxial tests. The results indicate that the algorithm is able to accurately predict the crack propagation under proportional, non-proportional, in-phase, and out-of-phase loading conditions.