International Journal of Fatigue ( IF 6 ) Pub Date : 2021-10-16 , DOI: 10.1016/j.ijfatigue.2021.106597 Lei He 1 , Zhilei Wang 2 , Yuki Ogawa 3 , Hiroyuki Akebono 4 , Atsushi Sugeta 4 , Yoshiichirou Hayashi 5
Fatigue tests were conducted under constant amplitude loading with a stress ratio of R = −1 to evaluate the effect of defect/inclusion on fatigue life of welding martensitic stainless steel (13Cr-5Ni) and KSFA90 steel (manufactured for crankshaft). An analysis based on the combined of defect/inclusion and linear elastic fracture mechanics (LEFM), was performed. The results indicate that defect/inclusion dominated the fracture mode for both materials utilized. Moreover, the combined and LEFM methods lose their accuracy for predicting fatigue life. Final, random forest machine-learning approach provided acceptable precision for unified predicting the fatigue of materials utilized in the current study.
中文翻译:
基于机器学习的缺陷/夹杂物对钢疲劳行为影响的研究
疲劳试验在恒幅载荷下进行,应力比为R = -1,以评估缺陷/夹杂物对焊接马氏体不锈钢(13Cr-5Ni)和 KSFA90 钢(为曲轴制造)疲劳寿命的影响。基于组合的分析缺陷/夹杂物和线弹性断裂力学 (LEFM),进行了。结果表明缺陷/夹杂物主导了两种材料的断裂模式。此外,联合和 LEFM 方法失去了预测疲劳寿命的准确性。最后,随机森林机器学习方法为统一预测当前研究中使用的材料的疲劳提供了可接受的精度。