Neural network approach to crack identification

Article type: Research Article

Authors: Sikora, R.^a | Chady, T.^a | Sikora, J.^{b; *}

Affiliations: [a] Technical University of Szczecin, Theoretical Electrotechnics and Computer Science Department, 70-310 Szczecin, Poland | [b] Warsaw University of Technology, Institute of the Theory of Electrical Engineering and Electrical Measurements, 00-661 Warsaw, Poland

Correspondence: [*] Corresponding author. Fax: + 48 22 292962; E-mail: [email protected].

Abstract: The eddy current non-destructive testing of conductive materials is a well-known problem. There are a lot of different kinds of transducers. For one of them the mathematical model was built and the Inverse Problem formulated. In order to solve this problem the Artificial Neural Network approach, relatively new, was adopted. The public domain software, developed at Warsaw University of Technology, was used to teach the network. It was proved that a standard feedforward multilayered neural network is sufficient to identify the parameters of the defect with a satisfying accuracy.

DOI: 10.3233/JAE-1998-119

Journal: International Journal of Applied Electromagnetics and Mechanics, vol. 9, no. 4, pp. 391-398, 1998