Characterisation of metal loss defects from magnetic flux leakage signals with discrete wavelet transform

Abstract

Metal loss defects in a buried pipeline are detected by magnetic flux leakage technique. Characterisation of the defects and sentencing according to the severity is extremely important for organised maintenance of pipelines. In this paper we identify the parameters that characterise a defect and the features of magnetic flux leakage signal (MFL) that are affected by those parameters. We show that analysis of the MFL signal using wavelet transform scores over any other method of its kind and exposes the incompleteness of the other analysis techniques that have appeared in the literature, to date. A number of experiments were performed on a rotating drum test rig having defects of different shapes and sizes. The results from these experiments are presented and discussed in detail. Wavelet transform decomposition and reconstruction techniques were applied for denoising the raw data. We test the efficacy of discrete wavelet transform for denoising MFL signal and present a complete scheme of characterisation of defects from denoised MFL signal. We discuss the issue of defect classification and suggest that characterisation to specified accuracy, amounts to designing a classifier that assigns a defect into known classes whose shapes and sizes are defined a priory.