PSI - Issue 60

Balaji Srinivasan et al. / Procedia Structural Integrity 60 (2024) 418–432 Balaji Srinivasan et al./ StructuralIntegrity Procedia 00 (2019) 000 – 000

423

6

separately in Figure 1 and Figure 2, respectively for clarity. For branch and header, the XGB model had shown a dominant performance compared to RFR model. From the MAE header and branch graphs, there is a significant difference between XGB and RFR models. The behavior is no longer consistent as in the case of MAE metrics. The higher the MAE value the lower the accuracy. The overall prediction for the branch is more accurate than the header, and XGB is more precise than the RFR model. However, for a few cases, like header outer surface out of plane loading prediction, the RFR model predicted better than the XGB model. For all 30 stress factors, the actual vs. predicted plot for the XGB and RFR models is presented in Appendix A. From the study of various input variables, the appropriate combination of XGB and RFR models is used to determine the stress factors in further studies.

R-Squared Error - Branch

R-Squared Error - Header

0.9840 0.9860 0.9880 0.9900 0.9920 0.9940 0.9960 0.9980 1.0000

0.9930 0.9940 0.9950 0.9960 0.9970 0.9980 0.9990 1.0000

B-M-Axial

H-M-Axial

B-M-Pressure B-Ins-Axial

H-M-Pressure H-Ins-Axial

B-Ins-Pressure B-Out-Axial

H-Ins-Pressure H-Out-Axial

B-M-Inplane

B-M-Torsion

H-M-Inplane

H-M-Torsion

B-Ins-Inplane

B-Ins-Torsion

H-Ins-Inplane

H-Ins-Torsion

B-Out-Inplane

B-M-OutPlane

B-Out-Torsion

H-Out-Inplane

H-M-OutPlane

H-Out-Torsion

B-Out-Pressure

B-Ins-OutPlane

H-Out-Pressure

H-Ins-OutPlane

B-Out-OutPlane

H-Out-OutPlane

XGB RF

XGB RF

Fig. 1. Comparison of Regression Models – R-squared Score of Branch and Header

MAE - Branch

MAE - Header

0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 0.6000 0.7000 0.8000

0.0000 1.0000 2.0000 3.0000 4.0000 5.0000 6.0000

H-M-Axial

B-M-Axial

H-M-Pressure H-Ins-Axial

B-M-Pressure B-Ins-Axial

H-Ins-Pressure H-Out-Axial

B-Ins-Pressure B-Out-Axial

H-M-Inplane

H-M-Torsion

B-M-Inplane

B-M-Torsion

H-Ins-Inplane

H-Ins-Torsion

B-Ins-Inplane

B-Ins-Torsion

H-Out-Inplane

H-M-OutPlane

H-Out-Torsion

B-Out-Inplane

B-M-OutPlane

B-Out-Torsion

H-Out-Pressure

H-Ins-OutPlane

B-Out-Pressure

B-Ins-OutPlane

H-Out-OutPlane

B-Out-OutPlane

XGB RF

XGB RF

Fig. 2. Comparison of Regression Models – MAE Score of Branch and Header

Table 5. R2 Error and MAE Comparison of XGB & RFR Stress Factors

R2 Error

MAE

R2 Error

MAE

No. Stress Factors Notations

No. Stress Factors Notations

XGB

RF

XGB

RF

XGB

RF

XGB

RF

1 2 3 4

B-M-Axial B-M-Inplane B-M-OutPlane B-M-Torsion

1.000 0.999 0.169 0.234 1.000 0.999 0.085 0.102 1.000 0.999 0.167 0.154 0.999 0.997 0.035 0.026

16 17 18 19

H-Ins-Axial H-Ins-Inplane H-InsOutPlane H-Ins-Torsion

1.000 0.998 1.090 1.374 1.000 0.999 0.430 0.396 0.999 0.999 1.555 0.969 0.999 0.997 0.183 0.153