A fast algorithm for 3D reconstruction of complex defect profiles for magnetic flux leakage inspection

In this paper, a fast and effective algorithm is proposed for the reconstruction of complex defect profiles for magnetic flux leakage (MFL) inspection, which is the most widely used non-destructive testing (NDT) technique for tanks and pipelines. A Gauss-Newton optimisation procedure is used to reconstruct the depth profile by optimising the parameters of the analytical model. The space mapping (SM) technique is employed to minimise the misalignment between the 'coarse' analytical model and the 'fine' finite element method (FEM) model. A region growing (RG)-based evaluation method is used to locate defect signals and estimate the opening shape of the defect, which reduces the number of optimisation parameters. To demonstrate the accuracy of the proposed inversion algorithm, the procedure is tested with defects of complex shapes from simulated MFL signals under different degrees of freedom (DOFs). The procedure is also tested using the experimental data of two metal loss defects. In all cases, the proposed algorithm shows good agreement between the actual and the reconstructed defect profiles, with fast convergence speed.