A novel data-driven auto compensation algorithm for pulsed eddy current inspection of high voltage feeder cable pipe

In this study, we introduce a novel auto-compensation method designed to enhance the accuracy of pulsed eddy current (PEC) measurements, crucial for non-destructive testing (NDT) in assessing the integrity of high-voltage feeder pipes. We explicitly target the unique challenges posed by strong electromagnetic interference (EMI) from internal power lines carrying three-phase alternating current. To address them, we propose a compensation algorithm that accounts for spatially varying magnetic permeability in the piping materials. The proposed method integrates multi-physics simulations of the interactions between the power lines, the pipe, and the PEC probe. Two finite element method (FEM) simulation models are developed: the Cable-Pipe Model, simulating the magnetic flux density distribution around the pipe due to internal power lines, and the PEC-Pipe Model, simulating the PEC sensing response considering the circumferentially varying magnetic permeability. To bridge the gap between computational accuracy and field deployment requirements, novel surrogate models are developed based on parametric FEM simulation datasets, allowing rapid approximation of characteristic decay time constants and nominal wall thickness estimations. Field tests validated our method's effectiveness, demonstrating a measurable reduction in false-positive indications after the physics-based compensation. This work challenges the conventional assumption of constant magnetic permeability in PEC inspections and extends the applicability of PEC for flaw detection in high-voltage feeder pipes and other challenging inspection scenarios, contributing to improved safety and infrastructure longevity.