Self-organizing adaptive neural fuzzy control for electrically driven robot manipulators

Abstract

A novel self-organizing adaptive neural fuzzy control (SANFC) is proposed for the trajectory tracking of an n -link robot manipulator including motor dynamics. In this control scheme, a self-organizing neural fuzzy network (SONFN) is constructed simultaneously to estimate online the system uncertainties with the structure and parameter learning phases. The fuzzy rules in the SONFN can be either generated or eliminated to obtain a suitable-sized network structure, and the adaptive tuning laws of network parameters are derived in the sense of the Lyapunov synthesis approach to ensure network convergence as well as stable control performance. A supervisory controller is used to attenuate the effects of the approximation error on the tracking error. The merits of the SANFC are that not only can the stable position tracking be achieved but also a suitable-sized network structure is found to avoid overfitting or underfitting data sets. Experiments performed on a two-link robot manipulator driven by direct current servomotors demonstrate the effectiveness of the scheme.