Page 57 - IJAMD-1-1

International Journal of AI
for Material and Design Integrating physics data for DL in DED

Table 5. Discretized parameter level for augmentation of 11, 13, and 19 were identified as outliers (Figure 15).
simulation data Furthermore, in the testing experimental dataset, samples
8, 15, and 32 were identified as outliers (Figure 16). In
Input parameters Levels summary, the outlier analysis pinpointed specific samples
LP (W) 1400 1500 1600 1700 1800 that deviated significantly from the majority within each
SS (mm/min) 2000 2250 2500 2750 3000 dataset – augmented simulation, simulation, experimental,
PMFR (g/min) 16 17 18 19 20 and testing experimental data. These deviations have
Sulfur content (wt%) 0.0005 0.004125 0.00775 0.011375 0.015 the potential to affect the performance and accuracy of
Abbreviations: LP: Laser power, PMFR: Powder mass flow rate, predictive models trained on this data. Therefore, careful
SS: Scanning speed. consideration is warranted to determine whether these

Figure 6. Data augmentation process adapted from Chen et al. .
32
Abbreviation: RSM: Response surface methodology.

Figure 7. Correlation analysis of augmented simulation data.

Volume 1 Issue 1 (2024) 51 https://doi.org/10.36922/ijamd.2355