Adaptive Control with Sliding Mode on a Double Fuzzy Rule Emulated Network Structure

Abstract

An adaptive controller based on neuro-fuzzy networks and sliding mode techniques is presented. The algorithm is intended for nonlinear discrete-time plants under the assumption that their mathematical model is unknown. Two adaptive Fuzzy Rule Emulated Network (FREN) structures are implemented to estimate the only two control parameters to be adjusted by a single neural network. The first FREN structure uses the error measurement as input. While the second one uses the sliding surface parameter to provide higher robustness. The stability analysis and performance of the proposed FREN Sliding Mode Controller (FRENSMC) for position control are presented in this work. The experimental setting consists of the tracking of a desired trajectory by controlling a DC motor along the X axis of a cartesian robotic system. The proposed FRENSMC controller achieves excellent results and exhibits a robust performance.