Ефективні оптимізаційні методи одночасного вимірювання профілів структурозалежних параметрів об’єктів засобами вихрострумового контролю

Abstract

In this paper we have carried out studies on determination of electrophysical parameters profiles of a cylindrical testing object by an eddy-current probe by proxy- optimization in the latent search space of reduced dimensionality. The peculiarity of these studies is the consideration of previously collected information on the regularities caused by variations of electrophysical parameters profiles of the testing objects. The functions of accumulator and carrier of apriori information are performed by the metamodel obtained on the basis of deep neural networks with high computational efficiency. Numerical experiments demonstrated the reliability of the proposed approach to determining the magnetic permeability and electrical conductivity distribution along the near-surface layers of metallic objects with microstructural changes. The analysis of the solutions results of the inverse measurement problem shows that the accuracy of profile reconstruction is quite high.