Intelligent real-time technique to estimate stress-corrosion of metal on the measured signals of eddy current flaw detector-scanner

Speaker:
Zagidulin, Timur; Spector Sci-Tech LLC; Russian Federation

Authors:
Zagidulin, R.V.; Spector Sci-Tech LLC; Russia
Zagidulin, T.R.; Spector Sci-Tech LLC; Russia

ID: ECNDT-0440-2018
Session: Data Processing - ET
Room: J2
Date: 2018-06-11
Time: 14:30 - 14:50

Practice of structural health monitoring on stress-corrosion estimation points to promise of multi-element eddy current flaw detectors-scanners with intelligent signal processing.

In general, the stress-corrosion may be considering as the group of defects of two basic types, arranged on the surface of metal: convenient cracks and pitting corrosion.

A physical model of eddy currents origin in the metal proposed, the consequence of which that is signals of eddy current transducer measured on cracks and on corrosion have significantly different shapes because they are describing by fundamentally different symbolic expressions depending on linear dimensions and shape of defect. On expressions obtained the two basic wavelets constructed, which allowed producing intelligent real-time locating of the defects on measured signal of eddy current transducer and separating them into two basic types: crack or corrosion of metal.

In order to estimate linear dimensions of defects detected the multi-parametric approach used to minimize smoothing functional composed from classification features of theoretical models of defects and features extracted from signals of eddy current transducer measured. Resulting solution of optimization problem set the linear dimensions of crack are determining by two parameters: its depth and width; at the same time, the linear dimensions of corrosion are determining by three parameters: depth, width and length of defect.

Increased accuracy of linear dimensions estimation of defects detected is reaching through the separate digital processing of eddy-current signals from cracks and corrosion, and increased number of informative features by coefficients of its wavelet transform.

Described technique of stress-corrosion estimation implemented in portable 32-channel eddy current flaw detector-scanner RSD «Nevod», executing signal processing on STM32F746 microcontroller in real time.

First applications of created eddy current flaw detectors in structural health monitoring of steel pipelines showed high accuracy, sensitivity and performance.