Fast defect parameter estimation based on magnetic flux leakage measurements with GMR sensors

Abstract

We present a fast inverse scheme that is capable of simultaneously estimating the parameters depth, opening and length for rectangular 3D geometries of surface-breaking defects. The parameter estimation is realized by an iterative least-squares minimization using the trust-region reflective algorithm. A semi-analytic magnetic dipole model that allows the sensor characteristics to be incorporated is used for predicting the stray magnetic fields. Giant magneto-resistance (GMR) measurements were carried out on a test specimen that includes a series of artificial defects. For the estimation of the defect depths relative errors between 0.6% and 15.9% have been obtained. Due to its very low computational costs, the inverse scheme can suitably be employed in automated production environments.