Process Modeling of Direct Ink Write 3D Printing Using Computer Vision and Machine Learning

Additive manufacturing (AM) is well-known for its ability to precisely place multiple materials at micrometer resolutions, with minimal limitations on structurally complexity, at a relatively low cost. However, a lack of confidence in AM for successfully producing high-quality, end-use parts has slowed its widespread implementation. This can be attributed to the complex and non-linear nature of AM processing parameters which are not well-understood by users and difficult to capture using traditional modeling methodologies. Therefore, this work seeks to model the complex relationship among process variables for direct ink write (DIW) 3D printing using machine learning (ML) algorithms. To do this, a big-data approach was implemented using computer vision techniques to monitor the DIW printing process in real-time. As a result, the complex relationship between the printed object and its corresponding DIW printing parameters were captured. The results provide an avenue for real-time, autonomous feedback control systems, establishing an AM framework which may improve both the quality and success rate of 3D printed components spanning a complex processing parameter space.