Accurate Identification of High Relative Density in Laser-Powder Bed Fusion Across Materials Using a Machine Learning Model with Dimensionless Parameters

Abstract

Machine learning (ML) methods have been extensively applied to optimize additive manufacturing (AM) process parameters. However, existing studies predominantly focus on the relationship between processing parameters and properties for specific alloys, thus limiting their applicability to a broader range of materials. To address this issue, dimensionless parameters, which can be easily calculated from simple analytical expressions, were used as inputs to construct an ML model for classifying the relative density in laser-powder bed fusion. The model was trained using data from four widely used alloys collected from literature. The accuracy and generalizability of the trained model were validated using two laser-powder bed fusion (L-PBF) high-entropy alloys that were not included in the training process. The results demonstrate that the accuracy scores for both cases exceed 0.8. Moreover, the simple dimensionless inputs in the present model can be calculated conveniently without numerical simulations, thereby facilitating the recommendation of process parameters.