Identification of flaws using eddy current testing

The eddy current method of non-destructive testing uses an alternating current excitation to induce secondary currents in a specimen under test. Flaws within the specimen affect the induced currents, causing changes in the impedance of a test coil. In this paper we present a method for obtaining a solution of inverse problems, in which the parameters of defects are unknown and the excitation function and the eddy current system response are given. The method is based on the use of artificial neural networks, which are trained using measurements. Illustrative examples are given.