Self-organizing non-linear dynamic object control system based on neuro-fuzzy networks

Abstract

The paper explores the creation of a self-organizing regulator utilizing a neuro-fuzzy network, capable of accurately approximating nonlinear functions with precision. Employing neuro-fuzzy networks as self-organizing regulators introduces nonlinear characteristics, extending the object's control range and enhancing adaptability within control systems. To streamline the learning process of the neuro-fuzzy network and ensure overall asymptotic stability, the proposal suggests subdividing the system model into smaller sub-models, effectively reducing dimensionality. This approach is not only beneficial for single-dimensional systems but also proves applicable to multidimensional control systems of nonlinear dynamic objects.