Information Based on Neural Network Modeling and Development and Quality Improvement of Management Systems

Abstract

Reliability and performance of information-control systems and their elements and nodes issues of quality improvement methods of identification and management of technological objects and networks and considered based on the models. Analytical of certain methods of control of neural networks were reviewed and given the advantages and disadvantages of each of them. A classification of control methods of neural networks was proposed. An imitative neural control circuit was analyzed, the main component of which is a neural network with the help of an existing regulator. Features of neural control, in particular, from two neural networks at the same time based on the usage, error passing through the feed back directly through the neuroemulator is described based on the method, and the second - as a direct neuroemulator acts as a regulator, Inverse error passing directly through the neuroemulator A method is presented for better control quality compared to previously considered methods provides.