Search for: All records

Abstract Direct ink writing (DIW) is an extrusion-based additive manufacturing technology. It has gained wide attentions in both industry and research because of its simple design and versatile platform. In electric-field-assisted Direct Ink Writing (eDIW) processes, an external electric field is added between the nozzle and the printing substrate to manipulate the ink-substrate wetting dynamics and therefore optimize the ink printability. eDIW was found effective in printing liquids that are typically difficult to print in the conventional DIW processes. In this paper, an eDIW process modeling system based on machine learning (ML) algorithms is developed. The system is found effective in predicting eDIW printing geometry under varied process parameter settings. Image processing approaches to collect experiment data are developed. Accuracies of different machine learning algorithms for predicting printing results and trace width are compared and discussed. The capabilities, applications and limitations of the presented machine learning-based modeling approach are presented.