Machine Learning-Enabled In-Situ Defect Detection in Extrusion-Based Additive Manufacturing: A Step Toward Qualification and Certification

Additive Manufacturing (AM), particularly Fused Deposition Modeling (FDM), is widely valued for rapid prototyping and cost-effective production. As a cornerstone of the Fourth Industrial Revolution (Industry 4.0), its integration with Smart Manufacturing principles including the Internet of Things (IoT), automation, and data analytics is essential for broader adoption. However, persistent print defects often result in material waste and part rejection, underscoring the need for advanced monitoring technologies. Such technologies form the basis of metrology (the science of measurement), which is critical to overcoming challenges in standardization and qualification that currently limit AM’s full potential. Building on prior work in multi-process monitoring and adaptive control, this study introduces a real-time, in-situ defect detection framework aimed at minimizing errors during printing. The system integrates an Ender V Pro 3D printer with an overhead optical camera and computational modules for continuous data processing. A machine learning-based image classification model is employed to detect defects in real time, enabling early anomaly identification, reducing filament waste, and minimizing manual intervention. The approach also evaluates material consumption and energy efficiency, demonstrating the value of predictive analytics in enhancing AM sustainability. By advancing in-situ monitoring, this system contributes toward effective Qualification and Certification (QC) of FDM components and marks a practical step toward fully autonomous AM workflows.