In the present work, for performing the ultrasonic flaw classification in weldments in a real-time fashion, an Intelligent Ultrasonic Evaluation System (IUES) is developed by the integration of four ingredients: 1) a PC-based ultrasonic testing (PC-UT) system, 2) database with abundant experimental ultrasonic flaw signals, 3) invariant ultrasonic pattern recognition algorithm using normalized features and probabilistic neural networks (PNNs), and 4) intelligent flaw classification software where the invariant algorithm is implemented. For the improvement of the classification performance, especially for slag inclusions, the feature set is enhanced by the addition of four informative features. In addition, a new effective feature selection scheme, the forward selection with feature evaluation criteria is proposed for the selection of subsets of sensitive features in a simple and straightforward manner without any trial and error.

This content is only available via PDF.
You do not currently have access to this content.