The identification and study of the critical state of a transverse crack in a zone with a lap based on artificial neural networks

• Published in: Russian journal of nondestructive testing 2014-08, Vol.50 (8), p.452-463
• Main Authors: Solov’ev, A. N., Sobol’, B. V., Krasnoshchekov, A. A.
• Format: Article
• Language: English
• Subjects: Acoustic Methods; Artificial neural networks; Characterization and Evaluation of Materials; Chemistry and Materials Science; Coating; Cracks; Digitization; Integral equations; Materials Science; Mathematical analysis; Structural Materials; Vectors (mathematics)
• ISSN: 1061-8309; 1608-3385
• DOI: 10.1134/S1061830914080087

A method for the identification of the geometry of a transverse crack in a layer with a lap with subsequent estimation of the critical state of its structure is proposed. The solution of the problem proceeds in two steps, which form the general method. The determination of a crack in a layer is performed with the use of artificial neural networks, which use the data of an ultrasonic study. To determine an internal crack, the problem of the pulse excitation of a wave field in a considered object was studied. Using a piezotransducer on a free coating surface, the results of an action are obtained. The input vectors for the artificial neural network are considered to be a regularly digitized echo signal. Based on the generated database of input vectors, the optimal training algorithm and the structure of the network were determined. After reconstruction of the geometry of a crack, the critical state is estimated with the use of integral equations, which were obtained from the solution of the corresponding linear problem of elasticity theory. In this problem, the coating medium is replaced by special boundary conditions on the upper surface of the layer. When using a generalized Fourier transform in the equilibrium equations in displacements, the problem is reduced to a singular first-order integral equation, whose solutions were derived by the small parameter and collocation methods. The stress-intensity coefficients were obtained and analyzed directly near the tops of the crack for various coating materials and geometric parameters of the crack.