Neural networks applications in mechanical systems

Abstract

The impossibility of the precise modeling of physical processes has demanded for the control of processes with unknown both by their structure and their parameters. High-quality control can be achieved only by a controlling system which will be able, during its work, to adjust its response according to the act of input and disturbing quantities. Neural networks appear to be one of these controlling systems. The possibility of their training can be considered using Lyapunov's stability theory. In this paper the preparation of the model of object (mechanical system) for neural networks control is pointed out.