The principle of prediction of complex time-dependent nonlinear problems using RNN

An approach based on recurrent neural networks (RNN s) is applied to verify the possibility of using surrogate models for the prediction of dynamic nonlinear problems. Modeling complex time dependencies is currently still a challenge, when the structure of the neural network needs to be adapted to the dynamics of the problem. In this paper, the possibility of using prediction in space-time problems is illustrated by the possibility of using it to predict the course of the current in a simple RL circuit that is powered by a voltage source.