Neuro-fuzzy Adaptive Control system for Discrete Dynamic Objects
Abstract
The paper proposes a methodology for creating an adaptive control system for a discrete dynamic object, containing a compensator and a regulator, made on the basis of the fuzzy Sugeno model and allowing to eliminate the side of disturbances and the regulation error. Synthesis algorithms for parametric and structural identification, which along with the back propagation method of error were used to adapt fuzzy models. The use of an adaptive identifier for a neuro-fuzzy control system of a nonlinear dynamic object is proposed, functioning under conditions of uncertainty changes in internal properties and the external environment. Real-time structural and parametric identification algorithms have been developed, which is a combination of the algorithm for identifying the coefficients of linear equations and the method of the theory of interactive adaptation.