Artificial neural network for parameter identifications for an elasto-plastic model of superconducting cable under cyclic loading

This paper presents an example of the use of an artificial neural network (ANN) for parameter identification of a theoretical model of the behaviour of a fibrous composite under transversal cyclic loading. A set of parameters of a generalised elasto-plastic model is identified to ensure the best accordance between two families of graphs of stress: that predicted by the theory and the experimental one. The adaptation of the theoretical model to obtain a better description of the experimental data is described in the paper. The application of the ANN technique for parameter identification is presented. An interpretation of the nature of mechanical processes that govern the analysed experiment is proposed and confirmed by the analysis of identified parameters.