In‐Situ Rheology Measurements via Machine‐Learning Enhanced Direct‐Ink‐Writing

Abstract

Direct ink writing, an extrusion‐based 3D printing method, is well suited for high‐mix low‐volume manufacturing. However, an iterative approach, using random selection or constant expert guidance, is still used to create printable inks and optimize printing parameters by expending significant amounts of time, materials, and effort. Herein, a machine learning (ML) model that estimates ink rheology in‐situ from a simple printed test pattern is reported. This ML model is trained with a rheologically diverse set of inks composed of different polymers. The model successfully correlated features of the simple printed test pattern to rheological properties, which could, in theory, inform both printed structures and future ink compositions. The behavior of this model is verified and analyzed with explainable artificial intelligence tools, linking printed feature importance to one's known physical understanding of the process.