Real-time in-situ thermal monitoring system and defect detection using deep learning applied to additive manufacturing

Abstract

Abstract. Fused deposition modeling, a widely employed additive manufacturing method, has witnessed a significant trend towards printing advanced materials such as PEEK and PAEK in recent years. Research studies have demonstrated the significance of process thermal dynamics in influencing the mechanical and geometric properties of printed components. This paper introduces a real-time thermal monitoring system that comprehensively tracks the thermal history of the printed component. Additionally, a deep learning model is presented, capable of detecting defects during the printing process. The integration of this monitoring system in a closed-loop mode offers the advantage of real-time adjustments, facilitating an immediate enhancement in the quality of the printed parts based on the continuously measured thermal data and the identified defects. Beyond real-time improvements, the data output from the monitoring system holds immense potential for broader applications. It can be seamlessly integrated into simulation software, providing a valuable dataset that can be leveraged to predict the physical properties and the adhesion quality of the printed parts.