Neurodynamic modeling of nonlinear non-stationary systems on the basis modern catasrophe theory

The problem of neural network modeling in the integrated environment of virtual testbed (VT) is discussed. Computer complex of the VT contains the interpreting and computing modules providing functioning neural-dynamic ND and graph-analysis of GA-systems. The ND-system contains neural-fuzzy NF and neural-evolution NE modeling. In aggregate ND and GA systems realize procedures of the control of dynamic situations in transport and social environments on the basis of functional a component of the modified system of the iterated functions (SIF). Management methods provide functioning of TS-system in the conditions of continuous change of dynamics of object and an environment. Adaptation is provided by means of the mechanisms of indistinct management defining a configuration « input - exit». Thus the feedback can be negative system at stabilization in pool attractor, or positive, conducting to bifurcation, to instability and chaos. The developed technology provides statement of problems of processing of the big data on high-lever subject-oriented language. The account of specificity of solved problems and semantics of a subject domain provides possibility of construction of the hybrid composit application combining various conceptual decisions. The structure of knowledge and the general approach to modeling of dynamics of TS-systems on the basis of the concept of subject-oriented language is considered. The formal device neural dynamic modeling at research of topological structures is realized on the basis of dynamic model of catastrophe. Topological dynamics of emergency calculations displays interpretation transport and social catastrophe. The approach to an estimation of intellectuality of computing technology is developed. The practical example of interpretation of «potential hole» is resulted at the analysis of the emergency connected with occurrence of «potential hole».