Machine learning predictions of superalloy microstructure

Gaussian process regression machine learning with a physically-informed kernel is used to model the phase compositions of nickel-base superalloys. The delivers good predictions for laboratory and commercial superalloys, R2>0.8 all but two components each γ γ′ phases, R2=0.924 (RMSE=0.063) fraction. For four benchmark SX-series alloys methodology predicts composition RMSE=0.006 fraction RMSE=0.020, superior 0.007 0.021 respectively from CALPHAD. Furthermore, unlike CALPHAD quantifies uncertainty in predictions, can be retrained as new data becomes available.