Automatic fault detection in Friction Stir Welding

Friction Stir Welding (FSW) is a relatively new welding technique where metals are joined through mechanical stirring. Due to its numerous advantages over older welding methods, it has been implemented in an increasing number of industries. However, there are remaining challenges to be overcome in FSW. One of the most serious is its reliance on accurate weld parameters. Additionally, faults or poor quality welds can develop from problems not easily detectible by an operator or robotic welder. In our work, we pursue automatic means of detecting fault occurrences and other quality problems. Force signals are collected from control welds run in aluminum as well as welds containing gap faults. Signal processing techniques, specifically Support Vector Machines (SVMs), are then used to correctly detect fault occurrences. Results demonstrate the ability for in-process fault detection of FSW.