Applying Machine Learning for Defect Detection in WAAM: A Literature Review

Additive manufacturing (AM) has rapidly advanced, offering unique advantages over traditional manufacturing, and is popular in research and industry. Due to its high deposition rates and cost-effectiveness, Wire Arc Additive Manufacturing is one of the AM processes that works best for making moderately to large-scale metal components. However, WAAM presents key challenges, such as inconsistent surface quality, process parameters, and thermal dynamics that affect dimensional accuracy and overall quality of the fabrication part. These issues complicate the control of geometric parameters like bead width, height, and penetration depth, often leading to deviations from required specifications. This review explores Machine learning (ML) applications in WAAM, which promise to address these production challenges. ML techniques support real-time process optimization, quality control, and predictive modeling in WAAM's complex manufacturing engineering and computational intelligence. ML models provide innovative, data-driven solutions to mitigate variability in WAAM processes, enhancing control and predictability. Machine learning is used for defect and result prediction in additive manufacturing. This study systematically identifies and evaluates key research findings, emphasizing the impact of ML on improving WAAM process outcomes. In conclusion, this review highlights the current challenges and emerging opportunities for advanced ML integration in WAAM.