Fig 1.
Schematic representation of the methodology used to evaluate the ML algorithms for predicting the relative density, surface roughness, and hardness of AlSi10Mg parts produced using SLM.
Table 1.
Chemical composition of AlSi10Mg [37].
Fig 2.
(a) SEM image of EOS AlSi10Mg powder, and (b) Gaussian distribution of the powder particle size showing mean and standard deviation [37].
Table 2.
Experimental conditions for relative density measurements.
Table 3.
Experimental conditions for the surface roughness measurements.
Table 4.
Experimental conditions for the hardness measurements.
Fig 3.
AlSi10Mg cube samples with the dimensions and printing direction fabricated using the EOS M400-4 AM machine.
Fig 4.
Mounted sample used for surface roughness and hardness measurements at five locations.
Table 5.
Details of the grinding and polishing steps used.
Fig 5.
Schematic of the ANN model.
Table 7.
Optimal tuning parameters and accuracy of the relative density, surface roughness and hardness models.
Table 6.
Selected process parameters and measured relative density, surface roughness, hardness for the 3D printed samples.
Fig 6.
(a) Optical microscope images showing porosities in sample 14 and 21, (b) mounted sample with location for hardness measurements and detailed measurements of sample 18, (c) SEM images of sample 14 and 25 showing the grain size distribution.
Fig 7.
Surface Topology of X-Y plane of SLM-processed AlSi10Mg for six different samples with their corresponding printing parameters.
Fig 8.
Predicted vs. actual relative density for
(a) ANN, (b) SVR, (c) KRR, (d) RF, and (e) Lasso based on LOOCV results.
Fig 9.
Predicted vs. actual surface roughness for
(a) ANN, (b) SVR, (c) KRR, (d) RF, and (e) Lasso based on LOOCV results.
Fig 10.
Predicted vs. actual hardness for
(a) ANN, (b) SVR, (c) KRR, (d) RF, and (e) Lasso based on LOOCV results.
Fig 11.
Relative density as a function of laser power and scan speed for AlSi10Mg parts using SLM
(a) 3D contour map and (b) 2D contour map with the optimized processing window shown in grey for relative density > 99%.
Fig 12.
Surface roughness as a function of laser power and scan speed for AlSi10Mg parts using SLM
(a) 3D contour map and (b) 2D contour map with the optimized processing window shown in grey for surface roughness < 10 µm.
Fig 13.
Hardness as a function of laser power and scan speed for AlSi10Mg parts using SLM
(a) 3D contour map and (b) 2D contour map with the optimized processing window shown in grey for hardness > 120 HV.
Fig 14.
Optimum Region for AlSi10Mg parts using SLM with the merged criteria of relative density > 99%, surface roughness < 10 µm, and hardness > 120 HV.