Flaw detection signals in the thermally insulated pipes are classified using artificial intelligence (AI). Nondestructive testing technology using pulsed eddy current (PEC) is a non-contact method for detecting surface flaws with high sensitivity; how...
Flaw detection signals in the thermally insulated pipes are classified using artificial intelligence (AI). Nondestructive testing technology using pulsed eddy current (PEC) is a non-contact method for detecting surface flaws with high sensitivity; however, the inspector is required to be highly skilled in signal reading. To replace the skill level of the inspector by AI, the measurement data were compared with the results of AI processing. A flaw with a depth of 50% of the thickness of the pipe was processed into diameters of 30, 60, and 90 mm in a 10-inch pipe, and these internal flaws were measured from the outside of the insulation material with a thickness of 50 mm using PEC. The PEC signals alone make it difficult to determine defect for flaws of 60 mm or less; however, AI processing data confirmed the presence or absence of defect. This study shows that even with seemingly meaningless experiments, useful results can be obtained by employing AI.