Fig.1

Algorithm flowchart for particle swarm optimization.

Table 1

Static performance parameters of 2014-T6

Fig.2

Stress-strain curve of 2014-T6

Table 2

Fatigue data for calibration of material parameters［31］

Table 3

Fatigue damage parameters of 2014-T6

Fig.3

Comparison between test data and calibrated fatigue model

Fig.4

Damage evolution numerical calculation flowchart

Fig.5

Variation of predicted error with ΔN/Nexp （Kt=1，R=0，σmax=344.75 MPa，Nexp =151 729）

Fig.6

Fatigue test specimen dimensional diagram of 2014-T6

Fig.7

Finite element model of 1/8 scale notched round bar specimen and schematic diagram of load and boundary conditions

Table 4

Sensitivity analysis of finite element mesh

Fig.8

Finite element analysis contour map

Fig.9

Variation of damage variable at critical point with the number of cycles

Fig.10

Variation of axial stress at critical point with the number of cycles

Fig.11

Axial stress-strain curve at critical point

Table 5

Calculation results of fatigue life based on finite element numerical method of damage mechanics

Fig.12

Comparison of experimental life and life predicted by numerical method of damage mechanics

Fig.13

Comparison of extended data， experimental data， and S-N curve

Table 6

SVM model parameters and performance evaluation indicators

Fig.14

Linear regression graphs of SVM model prediction results

Fig.15

Comparison of experimental life before and after correction with predicted life

Fig.16

Comparison of prediction error before and after correction

Fig.17

Damage variable vs number of cycles curve under different stress levels and damage variable vs number of cycles curve under different stress ratios

Fig.18

Damage variable curves with respect to cycle numbers under different stress concentration factors

Fig.19

Damage rate curves with respect to cycle numbers under different stress concentration factors

Fig.20

Linear regression graphs of test set with varying training data sizes

Fig.21

Variation of model prediction performance with sizes of training data

Fig.22

Comparison of predicted life and experimental life with varying training data sizes

Fig.23

Model prediction performance with variation in parameter C

Fig.24

Model prediction performance with variation in parameter γ