Monitoring Method of Friction Stir Welding (FSW) based on Tool Temperature Time-Series Data and Explainable Artificial Intelligence (XAI)

Friction stir welding (FSW), a solid-state welding process, is widely used in industries such as automotive and aerospace. Weld quality in FSW is primarily influenced by the frictional heat generated during the process. To more accurately capture this heat, this study measured the temperature of the rotating welding tool. Since the tool rotates during welding, a wireless measurement device was developed for real-time data acquisition. Al6061-T6 was used as the base material, and key features extracted from the temperature profile were applied as input features to an artificial neural network (ANN) model to predict tensile strength. Using explainable AI (XAI), the influence of each feature on model prediction output was analyzed, and process monitoring windows were established. The proposed method successfully identified FSW conditions likely to produce tensile strength outside acceptable limits. This study offers a practical approach for real-time quality prediction and monitoring in FSW, supporting the development of intelligent, data-driven, and autonomous manufacturing systems.