Noncontact geomagnetic defect localization of buried energy pipelines using ICEEMDAN approach with MVF

Abstract

Structural assessment of buried energy pipelines is often hindered by the abundance of external vibrations resulting in nebulous noises. Effective and secure nondestructive approaches need to be devised to efficiently reduce noise in multidimensional magnetic anomaly signals collected from a pipeline. This study focuses on the mechanism by which a measured source signal can be broken down into low- and high-frequency constituents known as intrinsic mode functions (IMFs). By doing so, a well-defined set of instantaneous frequencies is obtained utilizing improved complete ensemble empirical mode decomposition (ICEEMDAN) algorithms. These IMFs contain useful structural evidence across multiple scales that can be extracted for effective identification of the defect location. To accomplish this objective, first, the signal gradients are calculated using dual-density complex wavelet transform to diminish the influence of the geomagnetic field. The multiscale variance fusion (MVF) algorithm is then adopted to quantize the fluctuations occurring in each individual IMF. The output signals generated by computing the variances provide sufficient information about the location and severity of the pipeline defects. Numerical simulations for a buried pipeline model have been presented to validate the authenticity of the proposed technique. Indoor laboratory implantation on a pipeline test sample with prefabricated defects justifies the effectiveness of the ICEEMDAN-MVF model, to localize hidden structural flaws in energy pipelines without physical contact and even in more complex environments with multiple sources of magnetic interference.