•  

Algorithm of neural network method for creep model prameter identification problem

TitleAlgorithm of neural network method for creep model prameter identification problem
AuthorsE. B. Kuznetsov1, S. S. Leonov1, A. N. Vasilyev2
1Moscow Aviation Institute (National Research University)
2Peter the Great St. Petersburg Polytechnical University
AnnotationThe paper deals with a parameter identification problem for creep and fracture model that describe a process of metal structures deformation. The system of ordinary differential equations of Rabotnov's structural parameters kinetic creep theory is applied for describing this model. For solving the parameter identification problem, we proposed to use principals and techniques of neural network modeling. The procedure of neural network modeling application we are going to use for finding of uniaxial tension model parameters for isotropic steel 45 specimens at creep conditions. The obtained results of neural network modeling will be compared with theoretical strain-damage characteristics, experimental data and results of other authors.
Keywordscreep, fracture, damage parameter, artificial neural network, parameter identification problem, ordinary differential equation, initial value problem
CitationKuznetsov E. B., Leonov S. S., Vasilyev A. N. ''Algorithm of neural network method for creep model prameter identification problem'' [Electronic resource]. Proceedings of the XIII International scientific conference ''Differential equations and their applications in mathematical modeling''. (Saransk, July 12-16, 2017). Saransk: SVMO Publ, 2017. - pp. 97-100. Available at: http://conf.svmo.ru/files/deamm2017/papers/paper14.pdf. - Date of access: 24.06.2019.