CASPIAN JOURNAL

MANAGEMENT AND HIGH TECHNOLOGIES

Generalized procedure for the synthesis of algorithms neuronet identification based on theory entire functions

Read Zhashkova Tatiana V. Generalized procedure for the synthesis of algorithms neuronet identification based on theory entire functions // Caspian journal : management and high technologies. — 2013. — №4. — pp. 94-101.

Zhashkova Tatiana V. - Ph.D. (Engineering), Associate Professor, Penza State Technological University, 1a/11 Baydukov pr. / Gagarin St., Penza, 440039, Russian Federation, Zhashkovat@mail.ru

Purpose – to increase the efficiency of neural network identification of the complex systems on the monitoring parameters of the constituent physical objects. The methodological basis of the theory of the study was the identification, digital filtering, artificial neural networks (ANN), the mathematical theory of entire functions of exponential type (TSFET), and the methods of mathematical modeling and simulation. The author has solved the problem of the development of the analytical procedures of synthesis of systems of identification signal with respect to models based on the theory signaloobrazovaniya TSFET. The structure of the ANN, which feature is the presence in the first layer of special neurons dynamically storing the coordinates of the roots (zeros) of the signal. Performed numerical experiments showed that the probability of identification of polynomial signals using the proposed ANN is higher than for traditional ANN identifying signals at current values. This result is related to the fact that any entire function can be described by its roots (zeros).

Key words: analog-to-digital converter,information object,artificial neural network,complex system,entire function of potential type,digital signal processing