MLP Neural Network Observer Based RBF Neural Network Terminal Sliding Mode Controller Design For A Class Of Nonlinear System

This thesis presents an observer-based controller for a class of nonlinear systems design.In order to solve the issue of unknown dynamic in the practical systems,two distinctive neural networks(NNs)for observer and controller parts are applied.In the observer part,a multilayer perceptron(MLP)neural network based on the back-propagation(BP)error algorithm is adopted to update the weights of neural network that would be used to estimate the system states.In the controller part,radial basis function(RBF)neural network is combined with terminal sliding mode controller(TSMC),based on a new sliding surface.The update law in the controller part is related to the proposed sliding surface.The stability of the closed loop system is analyzed by Lyapunov stability theory.In order to verify the control performance of the proposed observer-based controller,it is compared with two other methods.In the first method,the sliding mode controller is applied,while in the second method the terminal sliding mode controller with a conventional sliding surface is used.The results of the simulation,applied on a chaotic system,show the superiority and better performance of the proposed approach by ensuring fast convergence,achieving a finite time stability,improving the robustness and eliminating the effects of disturbances.Furthermore,this work develops the design of the controller part by introducing a new fractional fast terminal sliding mode controller with radial basis function neural network.A new sliding surface is proposed to design the fractional fast terminal sliding mode controller,in order to obtain high precision and fast tracking performance.The suggested controller is used to synchronize different unknown master-slave dynamics with uncertainties.The neural network is applied to distinguish the unknown nonlinearity of both systems.The fast terminal sliding mode(FTSM)is applied to synchronize the systems in a small bounded time.In simulations,the proposed technique is applied on a chaotic system to demonstrate its effectiveness.