Data-Driven Optimization and Defect Prediction in WAAM of Nitinol Shape Memory Alloy

Wire Arc Additive Manufacturing (WAAM) offers significant advantages for fabricating large-scale metallic components due to its high deposition rate and material efficiency. However, the process remains highly sensitive to variations in process parameters, particularly when processing thermally and compositionally sensitive alloys such as Nitinol shape memory alloy. Conventional trial-and-error approaches for parameter selection often result in inconsistent quality, defect formation, and limited process repeatability. This study proposes a data-driven framework for process optimization and defect prediction in WAAM of Nitinol using machine learning techniques. Experimental data obtained from WAAM-fabricated Nitinol samples are utilized to develop predictive models correlating key process parameters with defect occurrence and mechanical performance. Multiple regression-based and machine learning models are evaluated to assess prediction accuracy and parameter sensitivity. The proposed approach demonstrates the potential of artificial intelligence to reduce defect probability, enhance process reliability, and support systematic optimization of WAAM processes for advanced manufacturing applications.