A Characteristic Approximation Approach to Defect Opening Profile Recognition in Magnetic Flux Leakage Detection

Defect opening profile recognition is a common problem in magnetic flux leakage (MFL) detection, while the traditional defect edge detection methods are not accurate enough. Starting from the basic principle of the electromagnetic field, this article discusses the model of the corresponding relationship between defect MFL signal and defect opening profile and analyzes the errors of traditional defect edge detection methods. Furthermore, the approximation characteristic of the MFL signal is proposed, and a characteristic approximation approach (CAA) is developed to detect the defect edge. The rectangular opening profile, circular opening profile, and arbitrary opening profile in FEM simulation and the experiments all proved the opening profile recognition capability of the proposed CAA. Compared with traditional defect edge detection methods, CAA reduces the error of profile recognition from 6.00% to 2.00%, which can facilitate the defect sizing in MFL detection. CAA also has high accuracy in the recognition of the tangential profile, which improves the capability of traditional MFL to detect tangential recognition.