Intelligent Control Of Nonlinear Systems With Unmodeled Dynamics

In the actual control system,due to the existence of modeling errors,external disturbances and other factors,the nonlinear system with unmodled dynamics is widely existed in the control system.The existence of unmodeled dynamics will affect the control performance of the closed-loop system,which leads to the insbility of the control system.Therefore,the controller design and performance of nonlinear systems with unmodeled dynamics have practical significance.This paper deals with several classes of nonlinear systems with unmodeled dynamics,by employing the backstepping technology and the adaptive neural network(fuzzy logic)control method,the controllers of these systems are designed,and the stability and the convergence problems of the corresponding closed-loop systems are analyzed.The main works are given as follows:The first chapter,for the robust fuzzy adaptive funnel control of nonlinear systems with dynamic uncertainties,Firstly,an improved funnel error is presented and merged into controller design to achieve performance bounds on tracking errors.By the utilization of approximation property of fuzzy logic systems and the backstepping design,an adaptive fuzzy funnel controller is constructed,which ensures that all signals in the closed-loop system are semiglobally uniformly ultimately bounded and the tracking error evolves within a pre-spectied performance funnel.Finally,simulation results are provided to demonstrate the proposed control approach.The second chapter,for the neural-network-based tracking control for a class of time-delay nonlinear systems with unmodeled dynamics,Firstly,the unknown nonlinearities are approximated by the radial basis functions(RBF)neural networks and the backstepping method is applied to present a systematical design framework.It is verified by the theoretical analysis and the simulation results that the presented controller guarantees the semi-global boundedness of all signals inthe closed-loop systems and the output tracking error eventually converges to a small area around zero.