DEEP LEARNING-BASED SURFACE AND INTERNAL DEFECT CLASSIFICATION IN METAL ADDITIVE MANUFACTURING

Abstract

Metal additive manufacturing has gained significant attention in advanced industrial applications due to its ability to fabricate complex components with high design flexibility and reduced material waste. However, surface and internal defects such as cracks, pores, lack of fusion, inclusions, and dimensional irregularities remain major challenges affecting component reliability and production quality. This paper proposes a deep learning-based defect classification framework for metal additive manufacturing by integrating convolutional neural networks, computer vision, Industrial Internet of Things (IIoT), thermal imaging, Digital Twin technology, multi-sensor data fusion, and predictive analytics. The proposed methodology analyzes surface images, thermal signatures, and process monitoring data to automatically classify manufacturing defects and support real-time quality assurance. Experimental evaluation demonstrates improvements in defect classification accuracy, inspection efficiency, process monitoring reliability, and manufacturing productivity. The proposed framework provides a scalable and intelligent solution for automated defect inspection and quality enhancement in Industry 4.0-enabled metal additive manufacturing environments. Keywords— Deep Learning, Metal Additive Manufacturing, Defect Classification, Computer Vision, CNN, IIoT, Digital Twin, Smart Manufacturing.