Abstract
Several industries extensively utilize polymeric foams due to their exceptional characteristics. The mechanical properties of a foam structure play a significant role in compression, so it is necessary to optimize foam deformation to achieve the desired outcome. The cell structures of foams are created randomly, but the issue has been resolved by lattice structures. Compared with traditional foams, lattice structures can enhance mechanical properties and facilitate the development of novel applications. Despite extensive research on lattice structures in both rigid and soft materials, there is a notable lack of predictive modeling specifically for soft thermoset Ultraviolet (UV)-curable lattice structures. This study employs additive manufacturing (AM) and machine learning (ML) to address this discrepancy. In this work, 93 lattice designs were produced using AM and evaluated for their geometric structure through compression tests utilizing ML techniques, specifically artificial neural network (ANN) and random forest (RF). The process involves the preparation of data, training of ML models, and evaluation. The RF model surpasses ANN model and is the most effective at predicting lattice geometries using force, strain, and lattice-type inputs in a graphical user interface. Hence, this study improves ML comprehension and utilization in the design of lattice structures to optimize the performance of soft materials across diverse applications.

This publication has 62 references indexed in Scilit: