Robust sliding mode control based on a new Quasi-sliding mode and adaptive artificial neural networks observer for robot

Abstract

This study designs and evaluates the simulation results of a robust sliding mode control based on a new Quasi–sliding mode and adaptive radial basis function neural network (RBFNN) observer applied to single-link robot control. An industrial robot (robot manipulator) is a multifunctional manipulator that can be programmed to perform dangerous and/or repetitive tasks with high precision. The robust adaptive RBFNN neural network observer is used to estimate the states and nonlinear functions in the mathematical description of the robot. The sliding mode controller based on a new Quasi–sliding mode combines with the robust adaptive RBFNN observer for the robot trajectory tracking control with appropriate quality indicators. The weights of the RBFNN are updated online. The stability of the proposed control methods is proven by Lyapunov stability theory. The simulation results in MATLAB/Simulink have shown the effectiveness and sustainability of the proposed method without the steady state error, the rising time achieves 0.4656(s), the settling time is 0.7690(s), the overshoot is 0(%), the values of RMSE (Root Mean Squared Error), IAE (Integral Absolute Error) and ISE (Integral Square Error) are 1.7549e-06, 0.0222 và 0.001124, respectively.