Khmelov A. Neurodynamic methods and models of identification of complex economic systems

Object of research: process of authentication of the difficult economic systems. Purpose of research: development of complex of neurondynamics models and methods of authentication of the difficult economic systems for an analysis, prognostication and optimization of business processes. Methods of research: methods of system-structural, comparative analysis and synthesis, methods of nonlinear dynamics, methods of mathematical analysis of temporal rows, methods of theory of automatic control, vehicle of artificial neuronnetworks, theory of evolutional algorithms, imitation modelling. Results: conception of neurondynamics authentication of the difficult economic systems with the use of vehicle of neuronnetworks; architectural decisions of dynamic neuronnetworks models; method of multidimensional decouplig of business processes of enterprise as a difficult economic system; method of formalization classification of informative channels of enterprise; neurondynamics models on the basis of associative neuronmashine; fitting for formalization of structural and self-reactance authentication of dynamic processes in the informative channels of the difficult economic systems; method of authentication of dynamic descriptions of the difficult economic systems on the basis of singular spectrology of temporal rows; models of prognostication and estimation of the state of key indexes of production and optimization of the system of budgeting of enterprise on their basis; complex of models of informative channels of enterprise at the direct design of their dynamic descriptions and for the construction of the generalized arrays of artificial educational selections. Novelty: First: conception of neurodynamics authentication of the difficult economic systems is worked out; the new evolutional methods of forming of neurodynamics models of the difficult economic systems are offered with the invariant size of genome of neural nets architecture; the model of classification of informative channels of enterprise is worked out and investigational by the nature dynamic business processes; the method of singular spectrology of temporal rows is offered with the neural nets search of компонент of decomposition on the basis of theory of factor analysis. Improved: neurodynamics models on the basis of associative neural nets; going near creation of structure of the distributed computer system of enterprise, to the calculable complication of processes of neural nets authentication intended for indemnification due to realization of parallel calculations within the framework of existent computer park; models of prognostication and estimation of the state of key indexes of production with periodic authentication of structure and parameters; methodological provisions of the basic stages of decouplig of business processes of enterprise as a difficult economic system on separate informative channels. Field of the use: level of enterprises.