Multi sensor probe and defect classification in eddy current tubing inspection

The eddy current (EC) technique is widely used for non-destructive evaluation of metallic structures because of its ease of implementation. However, EC signals relative to small defects are often corrupted by high magnitude perturbation signals, which make defect detection and/or characterisation difficult. In this paper a signal processing based on wavelet/maximum likelihood approach is presented and applied to the signals provided by an EC multi-coil probe dedicated to the testing of steam generator tubing. The proposed method shows good detection and classification performances for defects as small as 10% deep, 100 μm wide and 15 mm long notches.