Quantification of complex defect geometries based on leakage magnetic signals in pipelines

Abstract In the field of magnetic flux leakage detection, quantifying complex irregular defects constitutes an inverse problem based on magnetic flux leakage detection and represents a critical issue for practical engineering applications. Firstly, this paper processes the magnetic flux density modulus obtained from triaxial data at the defect location to extract an approximate opening contour of the defect. Subsequently, applying differential principles, complex defects are subdivided into sub-defects. By matching against a simulation signal library labeled with length and depth information, the simulated signal most similar to the actual signal is selected, and its labels are read. Integrating these simulated rectangular defects enables reconstruction of the complex defect. Experimental validation demonstrates an average length error (intercept between the scanning path and the complex defect’s opening contour) of approximately 12.34% and an average depth error of approximately 10.45%.