Page 84 - MSAM-1-1

Materials Science in Additive Manufacturing A ML model for AM PSP of Ti64

and EBSD feature inputs on model accuracy, five MP+SEM+EBSD+XRD). The rationale for this approach
different combinations of features were designed for is to understand the individual and interaction effects
training and testing. The first combination uses only of machining conditions, grain size, grain density, grain
machining parameters (MP) as the ML inputs; the second orientation, and residual stress on machining behavior.
combination considers machining parameters and EBSD As common in ML models, training and testing of all five
features (MP+EBSD); the third combination considers combinations of design features were repeated multiple times
machining parameters and residual stress (MP+XRD); (ML runs = 10), where randomly selected sets of training data
the fourth design considers machining parameters (80%) and test data (20%) were applied. ML results to predict
and SEM microstructure functions (MP+SEM); and specific cutting power during machining of Ti-6Al-4V AM
the last design integrates all the input features (All = surfaces using PBF are presented in Figure 7, where the Y-axis

Figure 7. RMSE values in both XGBoost and linear regression cross validation for different feature combinations.

A B C

D E

Figure 8. F-score in different input training models: (A) Machining parameters, (B) machining parameters with EBSD data, (C) machining parameters
with residual stress data, (D) machining parameters with SEM microstructure PCs, and (E) all features.

Volume 1 Issue 1 (2022) 12 https://doi.org/10.18063/msam.v1i1.6