Neural network modelling of evolutionary dynamics of nonlinear non-stationary system in the neighbourhood of «the potential well

The problem of neural network modeling of a nonlinear non-stationary system on the basis of the dynamic catastrophe theory is considered. Evolutionary dynamics of the system develops in the neighbourhood of the "potential well" under the influence of random perturbations. Behaviour patterns reflect the behaviour of the system under conditions of uncertainty and incompleteness of the initial information. The practical application of the developed conceptual solutions is discussed with reference to the interpretation of the interaction of a damaged ship in a continuously changing non-stationary dynamic environment. On the basis of the selected directions, principles and concepts, a mathematical apparatus for geometric and analytical interpretation of current situations has been developed in the complex use of the achievements of modern computer mathematics and high-performance computing tools.