SLIDING MODE CONTROL WITH RBF NEURAL NETWORK FOR TWO LINK ROBOT MANIPULATOR AND AN INVERTED PENDULUM SYSTEM

Abstract

Nonlinear control techniques are applied on two different mechanical systems namely two link robot manipulator and an inverted pendulum system to study the effect of the controllers on the tracking performance of the two system. A design of sliding mode control(SMC) for the position tracking of two link robot manipulator based on the sliding mode control technique and the Lyapunov stability theory is carried out to eliminate the perturbation and asymptotical stability can be achieved when the system is subjected to the sliding mode. A sliding mode control method based on RBF(radial basis function) neural network is addressed which has the capability of learning uncertain control actions shown by the several industrial robots. In RBFNN-SMC method the algorithm for tuning the parameters are extracted from the RBF function. The comparative study is done based on the evaluated parameters for the system. There are several practical applications like self-balancing robot, rocket propeller, Segway which are based on inverted pendulum. The control of inverted pendulum is carried out and a comparative analysis is made for the linear as well as the nonlinear model depending upon the two control techniques which are based on investigating the time, tracking error for obtaining the best performance for the inverted pendulum system. The implemented control techniques are the sliding mode control and the RBF neural network based adaptive sliding mode. Keeping the cart horizontal position and pendulum angle, to obtain the tracking performance the designed control law is subjected to different test signals and the results are shown for reduced chattering effect by using the above given controllers. A nonlinear SMC for position tracking control based on RBF neural network is also presented and the stability is given by the Lyapunov theorem.