A Method of Multi-crack Shape Identification from Eddy Current Testing Signals of Steam Generator Tubes Including Support Plates as Noise Sources

This paper deals with identifying multiple cracks from eddy current testing (ECT) signals obtained in a steam generator tube with a support plate and deposits. Assume two-dimensionally scanned ECT signals to be a picture image, then the signal processing by a multi-frequency technique eliminates noise caused by the support plate and deposits. A template matching with help of genetic algorithms detects number and positions of cracks from the image after the signal processing.Inverse analysis estimates the crack profile based on the predicted position of cracks. The number and positions of the cracks are sufficiently well predicted. Crack shape reconstructions are achieved with a satisfactory degree of accuracy.