N95-165 An Automated Real-Time Ultrasonic Weld inspection and Evaluation System

Weld inspection reliability is a major thrust area for the Navy's R&D program. Needs exist for nondestructive evaluation (NDE) technologies that can reliably detect, classify and size discontinuities in weldments, and that use this information to determine if the defects are acceptable or rejectable. Recent work performed by the Naval Surface Warfare Center with ultrasonic NDE methods have identified procedures and algorithms that address these needs. Independent research efforts, funded by the electric utility industry, have similarly indicated the potential of automating ultrasonic methods for the inspection and evaluation of pipe welds. Results from these programs are proposed to be used in this project to develop a real-time ultrasonic weld inspection and evaluation system. In Phase I, we will focus at advancing the state-of-the-art in two areas, namely: (1) the development of a sensor that can automatically track the position of an ultrasonic transducer, during manual scanning and (2) the development of a neural network based software package that can classify, size, and evaluate defects from ultrasonic NDE images. Feasibility studies performed in these areas should pave the way for the development of a system that can be used with manual and automated scanning techniques for weld inspection and evaluation.