Expanding the frontiers of additive manufacturing: Higher microstructure identification through probability modeling

Abstract

Probabilistic models and machine learning methods create a step forward in making predictions for additive manufacturing (AM) microstructure. In this probabilistic framework, it became possible to express modifications in the properties of metal, polymer, ceramic, and composite microstructures. Process parameters and material consistency reached maximum levels through the use of statistical modeling along with finite element analysis (FEA) and Gaussian process regression (GPR). Experimental validation through AM process parameters, microstructural values, and material characteristics led to 40 % fewer metal and polymer microstructure variations with simultaneous strength increases. The computational system demonstrated its resistance to process modifications through a validated sensitivity analysis. Additionally covered were scalability issues, computing needs, and possible real-time adaption. These results help AM approaches in aerospace and biomedical engineering to be scalable and performable.