PHYSICALLY MODELED FEATURE-BASED ULTRASONIC SYSTEM FOR IGSCC CLASSIFICATION.

This paper discusses the considerations that must be taken into account for the reliable detection and classification of intergranular stress corrosion cracks (IGSCC). Because 304 grade stainless steel (304SS) is the host material for IGSCC in this investigation, signed enhancement is critical. Good signals indicate only a reflector, but not its type. Reflector classification is explored using the methods of pattern recognition. Search unit requirements, based on acceptability for use and the possibility of compensating the unit for proper use, are discussed next. Decision schemes, or algorithms for classifying IGSCC versus geometric type reflectors, are explained along with the data base requirement necessary for system credibility. A description is given of possible instrumentation for such algorithms. Suggestions that involve the man-machine interface of in-service inspection intelligence and computerized intelligence are given for field usage. Lastly, the importance of good physical models of IGSCC is stressed.