Intelligent System For Adapting Typical Laws Of Automatic Control Based On Neural Networks

Abstract

The structure of an intelligent system for adapting typical automatic control laws with the function of identifying parameters and the structure of an object has been developed and described. The developed system solves a complex of interrelated tasks: identification of the structure and parameters of the object, setting and selection of the controller type, identification of parametric disturbances. The intelligent system consists of three levels: the identification level, the control system design level, and the executive level. The tasks of the identification level include: identification of the parameters and structure of the object by observing the transient response of the object with a step action at the input, identification of a parametric disturbance by the method of comparative assessment of the transient processes of the reference model and the real control object. As a result of identifying the parameters and the structure of the object, the controller is designed - setting the parameters and choosing the type of controller. The executive circuit is rearranged by a closed control system. Neural networks were chosen as a tool for implementing the system for identifying parameters and the structure of an object, for designing a control system. As a prototype of the control object, typical dynamic links and their connections were chosen. The article provides a general description of the structure of an intelligent system for adapting standard control laws, a description of individual functions, structural and graphical diagrams, and simulation results.