APPLICATION OF NEURAL NETWORKS FOR AUTOMATED INSPECTION OF NUCLEAR POWER PLANT PIPING

THE INSPECTION OF BUTT-WELDED STAINLESS STEEL PIPE JOINTS IN NUCLEAR POWERPLANTS IS ROUTINELY PERFORMED USING ULTRASONIC NON-DESTRUCTIVE EVALUATION METHODS. FIELD EXPERIENCE, BASED ON CONVENTIONAL ULTRASONIC SIGNAL AMPLITUDE CRITERIA, SHOWS THAT A LARGE NUMBER OF INDICATIONSARE RECORDED. MOST OF THESE ARE NOT CAUSED BY CRACKS, BUT THE WELD GEOMETRY AND IMPEDANCE MISMATCH BETWEEN THE WELD AND THE BASE MATERIAL. CURRENTLY, DISCRIMINATION BETWEEN CRACK AND GEOMETRIC/WELD (MALIGNANT VERSUS BENIGN) INDICATIONS IS BASED PRINCIPALLY ON OPERATOR EXPERIENCE, VARIATIONS IN SIGNAL AMPLITUDE, AND THE LOCATION OF THE REFLECTOR. SIGNIFICANT DIFFERENCES IN PERFORMANCE ARE CAUSED MOSTLY BY OPERATOR EXPERIENCE, FATIGUE, CONCENTRATION, AND CONVENTIONAL SIGNAL AMPLITUDE EVALUATION CRITERIA. IN ADDITION, THE PROCESS OF DISTINGUISHING THE TYPE OF REFLECTOR IS VERY TIME CONSUMING, AS FIELD EXPERIENCE AND ROUND-ROBIN TESTS HAVE SHOWN. IN RESPONSE TOTHIS INSPECTION PROBLEM, PATTERN RECOGNITION METHODOLOGIES AND IMAGING TECHNIQUES HAVE BEEN DEVELOPED IN THE PAST DECADE. HOWEVER, BOTH NEED CONSIDERABLE IMPROVEMENT TO ENSURE SAFETY AND RELIABILITY OF PRIMARY PIPING SYSTEMS. NEURAL NETWORKS THUS HOLD TREMENDOUS POTENTIAL FOR SOLVING CLASSIFICATION PROBLEMS. IN THIS PROJECT, THEY WILL BE APPLIED TO AUTOMATICALLY DETECT AND CLASSIFY THE INTERGRANULAR STRESS CORROSION CRACKING IN AUSTENITIC STAINLESS STEEL PIPING.