Sensor-integrated data acquisition and machine learning implementation for process control and defect detection in wire arc-based metal additive manufacturing

Abstract

Wire Arc Additive Manufacturing (WAAM) offers significant advantages, such as a high deposition rate and cost-effectiveness compared to conventional manufacturing methods. However, WAAM faces challenges related to porosity, cracking, inclusions, lack of fusion, and geometric inaccuracies, which degrade component quality and performance. To optimize the quality of the built parts and minimize the potential defects, a good approach is to use sensors-based monitoring and controlling the process. In the process the date can be recorded by using sensors and send real-time feedback to the control system. This study aims to provide a comprehensive review of in-process sensing technologies and their integration with control systems to mitigate these defects. This work systematically categorizes sensing approaches including optical, acoustic, visual, thermal, and multi-signal methods while emphasizing the role of machine learning in real-time data processing for defect detection and control. Effectively detecting different types of defects usually requires various sensing technologies, specific focus areas, and careful attention. This also involves real-time data integration and advanced data processing. A key novelty of this review lies in its critical evaluation of multi-sensor integration strategies, real-time data fusion techniques, and their potential to enhance WAAM process reliability. By addressing the fundamental principles, current limitations, and future research directions, this study serves as a valuable resource for advancing intelligent monitoring solutions in WAAM.