Towards automated interpretation of ultrasonic NDT data

Expert systems have demonstrated potential for automated interpretation of ultrasonic NDT data, but rules do not provide the best solutions for all in terpretation problems. In this work, 3D images are reconstructed from raw A -Scans, using CAD models of test specimen geometry to correct positions and remove geometric echoes. A knowledge model for ultrasonic NDT has been imp lemented in a working expert system, with uncertainty managed using fuzzy l ogic. A suite of data processing tools is available both to the expert syst em, and also through an interactive 3D visualization system, which therefor e provides a development environment for automated interpretation, as well as being powerful in its own right. (C) 1998 Elsevier Science Ltd. All righ ts reserved.