Issue 72

X. Cao et alii, Frattura ed Integrità Strutturale, 72 (2025) 162-178; DOI: 10.3221/IGF-ESIS.72.12

= = ∑ m m ME Error m 1 1

(19)

2 is the experimental cycle ratio under stress σ 2 , m f n N 2 2

n N

2

is the predicted cycle ratio under stress σ 2 , and m is

where

f

the sample size of the dataset. The smaller the values of Error and ME mean that the model performs better. Aluminum alloy welded joints The experimental data utilized for the tests comprised the fatigue test results of welded joints from the aluminum alloy bodies of high-speed trains in papers [20,21]. The welded joints used in the test contain two types of welding: butt joint and corner joint. The samples' base material is ENAW6005, an aluminum alloy employed in the construction of CRH2 high speed train bodies. Based on fatigue testing outcomes, the butt joint fatigue life of 549,300, 880,500, and 1,540,100 cycles at stress levels of 104, 89, and 74 MPa, respectively. Similarly, the corner joint's fatigue life were 619,800, 952,300, and 1,546,100 cycles at stress levels of 93, 83, and 73 MPa, respectively. The error between experimental and predicted values for butt joint and corner joint are shown in Fig 4 and 5.

(A) Butt joint without data augmentation (B) Butt joint with data augmentation Figure 4: Comparison of predicted and experimental cycle ratios for aluminum butt joint.

(A) Corner joint without data augmentation

(B) Corner joint with data augmentation

Figure 5: Comparison of predicted and experimental cycle ratios for aluminum corner joint.

172