Abstract The focus of this study is to predict curved crack FCG failure under variable amplitude load effectively and accurately. Based on the artificial neural network (ANN) and FCG path/life prediction models, a numerical calculation method is designed. The proposed method considers the underlying physical mechanism of cracked structure, and only a relatively small amount of finite element calculations are required to predict FCG problems with different initial conditions. Furthermore, it can be found that the geometric parameters of hole and crack have an effect on FCG. Finally, compared with the experimental and simulation results of different examples, the effectiveness of the new method is verified.

中文翻译：

---

基于人工神经网络的变幅载荷下弯曲疲劳裂纹扩展预测

摘要 本研究的重点是有效、准确地预测变幅载荷作用下的弯曲裂纹FCG失效。基于人工神经网络(ANN)和FCG路径/寿命预测模型，设计了一种数值计算方法。所提出的方法考虑了裂纹结构的潜在物理机制，并且只需要相对少量的有限元计算来预测不同初始条件下的 FCG 问题。此外，可以发现孔和裂纹的几何参数对 FCG 有影响。最后，与不同算例的实验和仿真结果进行对比，验证了新方法的有效性。