3-D Defect Profile Reconstruction from Magnetic Flux Leakage Signals in Pipeline Inspection Using a Hybrid Inversion Method

In this paper, we propose a hybrid inversion approach to reconstruct the profile of arbitrary three-dimensional (3-D) defect from magnetic flux leakage (MFL) signals in pipeline inspection. The region of pipe wall immediately around the defect is represented by an array of partial cylinder cells, and a reduced forward FE model is developed to predict MFL signals for any given defect. The neural network (NN) method is used at first to give a coarse prediction of the defect profile, and the prediction is then utilized as one original solution of the genetic algorithm (GA) to search for the global optimum estimate of the defect profile. To demonstrate the accuracy and efficiency of the proposed inversion technique, we reconstruct defects from both simulated and experimental MFL signals. In both cases, reconstruction results indicate that the hybrid inversion method is rather effective in view of both efficiency and accuracy.