Machine Learning Enabled Smart Structural Materials Using Additive Manufacturing

ABSTRACT: This research study describes a machine learning (ML)-driven model for producing smart structural materials via additive manufacturing (AM) by extrusion. A 3D concrete printing system was used to make cementitious composites that were reinforced with carbon nanotubes (CNTs) and graphene nanoplatelets (GNPs). Random Forest (RF), Support Vector Machine (SVM), and Artificial Neural Network (ANN) models were used to undergo supervised learning on an experimental dataset consisting of 320 specimens to predict compressive strength, electrical conductivity, and print quality as dependent on process parameters and material composition. The highest R² of compressive strength prediction of SVM was 0.946, whereas RF had the highest R² of 0.987, which was used to predict electrical conductivity. Optimization of parameters guided by ML had a 61.8% enhancement of compressive strength and 30.5 times increase in electrical conductivity in comparison to non-optimized baselines. Nanomaterial networks were also found to be conductive, allowing individual networks to detect their strain levels through changes in current at a strain of 0.1%, which facilitates real-time structural health sensing. The artificial system showed a 31% decrease in CO₂ emissions and a 58.8% decrease in material wastage compared with the usual way of building, proving to be a valid route towards intelligent and sustainable infrastructure.