Abstract
This paper presents a methodology for classifying the common defects in steel pipelines for transporting petroleum and gas. Usually, the nondestructive evaluation (NDE) experts in the industry judges the defect type by mere observation which, based on the experience, may or may not be correct. The proposed methodology has attempted to map this heuristic understanding from the shape of the defect waveforms (A-scans) using ultrasonic sensors with the help of fuzzy logic and fuzzy set associations. As such, a subset of features was selected for a set of commonly occurring defects and a fuzzy inference system is then generated using heuristic rules to classify the defect. The initial tests have shown over 90% success rate which is promising for further investigation.