Support vector regression-assisted finite element method for mode I-II fatigue crack growth path prediction

Abstract

Most of failures in mechanical components in engineering are caused by fatigue crack growth (FCG), especially mode I-II cracks. Accurately and fast predict of crack growth paths of mode I-II in structures is important in academic research and engineering applications. Finite element method (FEM) simulation of FCG can be accurately performed for engineering applications. However, the commercial FEM simulation software, e.g., ABAQUS combined with FRANC3D is generally time-consuming. Therefore, an accurately and fast mode I-II crack growth path prediction method is developed by support vector regression (SVR)-assisted FEM analysis. Firstly, the initial location data of FCG path and stress intensity factor (SIFs) are collected by FEM simulations. The SVR model is trained using initial location data and the corresponding FCG path will be predicted for several growth steps. Secondly, the predicted crack results are input into the FEM simulation model to obtain the several other growth steps. Finally, the two procedures are repeat alternately to obtain the full FCG path. The numerical case is given to validate the performance of the present method using the double-hole plates with mode I-II crack growths. Comparisons show that the performance of the present method are superior to those of FEM simulation and SVR alone.