Prediction of the fatigue curve of high-strength steel resistance spot welding joints by finite element analysis and machine learning
Zhengxiao Yu, Ninshu Ma, Hidekazu Murakawa, Goro Watanabe, Mingyao Liu, Yunwu Ma
Abstract
Abstract The process of resistance spot welding is extensively utilized in automotive assembly. Analyzing the fatigue strength of resistance spot welded (RSW) joints of thin plate high-strength steel holds significant importance in reducing production costs and enhancing vehicle safety during operation. By combining finite element analysis (FEA) and machine learning (ML), a novel method has been developed to predict fatigue curves of RSW joints with high-strength steels of different thicknesses, widths, and nugget diameters. In this study, the impact of various experimental conditions, such as the thickness and width of the sheet material, and the diameter of the nugget, on the fatigue test results, has been quantified. Moreover, the model established through this research enables accurate prediction of the F-N fatigue curves of RSW joints without the need for fatigue testing, thereby saving costs and time required for experimentation. The average error is approximately 8% of the experimental results.