Fatigue life prediction through defects is a challenging work. An approach to predict fatigue life based on defects information was proposed. Defects simplification and classification method was proposed based on damage data obtained by X-ray CT. Stress concentration factors was introduced as intermediate variable for virtual 3D reconstruction model of internal defects. A prediction model of relationship between defects and fatigue life was established by Support Vector Regression. It is shown that the proposed prediction model can built an accurate bridge between mesoscopic damage and macroscopic fracture by simple macro variables based on comprehensive defects information.

通过缺陷预测疲劳寿命是一项具有挑战性的工作。提出了一种基于缺陷信息的疲劳寿命预测方法。基于X射线CT获得的损伤数据，提出了缺陷简化和分类方法。引入应力集中因子作为内部缺陷虚拟3D重建模型的中间变量。通过支持向量回归建立了缺陷与疲劳寿命之间关系的预测模型。结果表明，所提出的预测模型可以基于综合缺陷信息，利用简单的宏观变量，在介观损伤和宏观断裂之间建立准确的桥梁。