| Abstract Scope |
Metal additive manufacturing (MAM) offers transformative potential for high-throughput, cost-efficient fabrication of complex metallic structures. However, its industrial adoption remains constrained by the absence of robust, real-time, in-situ quality assurance methodologies. Currently, defects such as porosity, distortion, and incomplete fusion are detected only after part completion, using expensive and time-consuming ex-situ techniques, such as X-ray computed tomography (XCT). This work presents a machine learning-enabled process fingerprinting framework that fuses multimodal in-situ sensors and ex-situ analysis data to extract quantitative process signatures for real-time process monitoring. Specifically, we analyze acoustic, thermal, and optical signals generated during the MAM process, employing multimodal data fusion to correlate dynamic process states with defect formation mechanisms. The proposed intelligent process fingerprinting tool lays the foundation for developing a digital twin capable of real-time feedback and adaptive process control, thereby enabling proactive defect mitigation and improved part quality in MAM. |