Deep Neural Network (DNN) Architecture
PyTorch / TensorFlow
Multi-Layer Perceptron (MLP) mapping LPBF Process Parameters to Microstructural Features.
Training Metrics
Epoch 142/200
Loss: 0.0342
Acc: 96.4%
Predictive Inference Engine
Live API
Adjust LPBF parameters to predict relative density and porosity using the trained surrogate model.
Laser Power (W)200
Scan Speed (mm/s)800
Hatch Spacing (μm)110
Predicted Relative Density
99.4%
Confidence Interval: ±0.12%
Active Training Datasets
Pandas / NumPy
Dataset NameSizeStatus
LPBF_InSitu_MeltPool_Thermal.csv
2.4 GB
Cleaned
NiTi_SMA_Composition_Phase_Map.json
450 MB
Cleaned
TPMS_Lattice_Stiffness_FEA.h5
1.1 GB
Augmenting
CT_Scan_Porosity_Defects_V3.dicom
8.5 GB
Queued
CFRP_Tensile_Failure_AE.wav
320 MB
Cleaned