On-line genetic fuzzy-neural sliding mode controller design

Abstract

In this paper, a novel online B-spline membership function (BMF) fuzzy-neural sliding mode controller trained by an adaptive bound reduced-form genetic algorithm (ABRGA) is developed to guarantee robust stability and tracking performance for robot manipulators with uncertainties and external disturbances. The general sliding manifold is used to construct the sliding surface and reduce the chattering of the control signal, which can be nonlinear or time varying. For the purpose of identification, the proposed BMF fuzzy-neural network trained by the ABRGA approximates the regressor of the manipulator. An adaptive bound algorithm is used to aid and speed up the searching speed of the RGA. Simulation results show that the proposed on-line ABRGA-based BMF fuzzy-neural sliding mode controller is effective and yields superior tracking performance for robot manipulators.