Neural network based chattering free sliding mode control

Abstract

We present a design method of online estimator which estimates a part of equivalent control containing a system's nonlinear term, input-matrix uncertainty and unknown disturbance by use of one of a neural network's most powerful ability, that is, function approximation, The controller using that estimated equivalent control is designed to have continuous control to eliminate chattering and to provide sliding mode motion on the selected manifolds in the state space.