Font Size: 

Modeling and understanding the underlying process-structure-property (P-S-P) relationshipduring additive manufacturing (AM) is vital to efficient process optimization and qualitycontrol. With the advancement of machine learning (ML) models and increasing availabilityof AM-related digital data, ML-based data-driven modeling has recently emerged as apromising approach towards exhaustively exploring and fully understanding AM P-S-Prelationship. Nonetheless, many of existing ML-based AM modeling severely under-utilizethe powerful ML models by using them as simple regression tools, and largely neglect theirdistinct advantage in explicitly handling complex-data (e.g., image and sequence) involveddata-driven modeling problems and other versatilities. To further explore and unlock thetremendous potential of ML, this research aims to attack two significant research problems:(1) from the data or pre-data-driven-modeling aspect: can we use ML to improve AM datavia ML-assisted data collection, processing and acquirement? (2) from the data drivenmodeling aspect: can we use ML to build more capable data-driven models, which can act asa full (or maximum) substitute of physics-based models for high-level AM modeling or evenrealistic AM simulation? To adequately address the above questions, a variety of ML models,including Gaussian process (GP), multilayer perceptron (MLP), convolutional neural network(ConvNet), recurrent neural network (RNN) and their variants, are leveraged to handlerepresentative data-driven modeling problems with different quantities of interest (QoI). Theyinclude data-driven process modeling (melt pool, temperature field), structure modeling(porosity structure) and property modeling (stress field, stress-strain curve).