Feature Engineering of Time-Domain Signals Based on Principal Component Analysis for Rebar Corrosion Assessment Using Pulse Eddy Current
Durgesh Tamhane, Jinit Patil, Sauvik Banerjee, Siddharth Tallur
Abstract
Corrosion of steel rebars in reinforced concrete impacts structural integrity of the infrastructure. While several Non-Destructive Testing and Evaluation (NDT&E) techniques exist for corrosion monitoring in structures, there is a need for cost-effective solution for early detection of sub-surface corrosion of rebars in civil infrastructure. Here we present a low-cost corrosion sensing system based on pulsed eddy current detection using anisotropic magnetoresistive (AMR) sensor to detect the corrosion in steel reinforcing bars. A novel probe with large figure of merit ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$21.62\times6$ </tex-math></inline-formula> m <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\text{T}^{-1}$ </tex-math></inline-formula> ) is presented, that can detect corrosion in 20 mm diameter rebars at lift-off distances as high as 55 mm with low excitation current of 0.2 A. The sensitivity is enhanced by using area under the curve of various segments of the time-domain sensor output as features. Optimal feature selection and dimensionality reduction, as well as quantitative estimation of extent of corrosion are demonstrated using principal component analysis.