EMPIRICAL LEARNING INVESTIGATIONS OF THE STRESS-CORROSION CRACKING OFAUSTENITIC STAINLESS-STEELS IN HIGH-TEMPERATURE AQUEOUS ENVIRONMENTS

A collection of data documenting the stress corrosion cracking (SCC) b ehavior of austenitic stainless steels (SS) in high-temperature aqueou s environments was investigated using empirical learning techniques. C omputer-based empirical learning systems based on classical and nonpar ametric statistics, connectionist models, machine learning methods, an d fuzzy logic are described. An original method for inducing fuzzy rul es from input-output data is presented. Performance comparisons of the various approaches are summarized, along with the relative intelligib ility of the outputs. In both respects, the decision-tree approach was found to perform very well on the problem investigated.