The Use of Feedback Forces for Nondestructive Evaluation of Friction Stir Welding
Author | : Enkhsaikhan Boldsaikhan |
Publisher | : |
Total Pages | : 288 |
Release | : 2008 |
ISBN-10 | : OCLC:308689759 |
ISBN-13 | : |
Rating | : 4/5 (59 Downloads) |
Book excerpt: The Friction Stir Welding (FSW) process provides a wealth of feedback information which can be used for evaluating weld quality in a nondestructive manner. The focus of this research is to perform a feasibility study for detecting wormhole defects using the frequency spectrum, phase space, and time spectrum of feedback forces in real-time. It is asserted here that the oscillations of the feedback forces are related to the dynamics of the plasticized material flow and that the feedback forces can provide useful information for detecting wormhole defects. Two real-time feedback-force evaluation algorithms are developed for assessing the weld quality. The first evaluation algorithm uses a neural network and the discrete Fourier transform. Overall, about 95% classification accuracy is achieved by the first algorithm. The second evaluation algorithm employs phase space approaches, which correctly identify micro-wormhole defects, defects that are less the 0.1 millimeter (mm) in diameter, with 82.8% accuracy. A probability of detection analysis is conducted to compare the wormhole-detection performance of the two algorithms. The phase space algorithm demonstrates slightly better wormhole detection probability that the neural network approach. However, both algorithms detect defects greater than 0.5 mm with 95% mean probability of detection, and both algorithms evaluate 3-second-long time series segments in less that 0.1 seconds (s). The detection probabilities are competitive with current nondestructive evaluation technologies, and the time taken for each process is small enough to allow this processing to be done in real time during welding. Thus, both algorithms are suitable for weld quality evaluation during welding which eliminates the need for post-weld processing. The present study not only demonstrates the possibility of identifying weld quality in real-time but also provides evidence that the oscillations of the feedback forces are related to the material flow. The study introduces a real-time approach for providing important feedback information about weld quality to a control system for FSW.