Intelligent Control For Classes Of Uncertain Nonlinear Systems

In practice, most control systems are nonlinear and often affected by uncertainties, delays and stochastic disturbance. As adaptive control method has the ability to identify object and adjust adaptive parameters online, it can be used to control uncertain systems. In addition, as fuzzy logic system and neural network can uniformly approximate an arbitrary continuous function to a given accuracy, they are effective methods to control uncertain nonlinear systems. In recent years, By combing backstepping design method and fuzzy logic systems or neural network, adaptive intelligent control method has been fully developed and many important results have been obtained. However, there still exist some open issues, which need to be further investigated. Therefore, intelligent control for classes of uncertain nonlinear systems, such as strict-feedback uncertain nonlinear systems, stochastic nonlinear systems and inter-connected large-scale nonlinear systems is investigated in this doctoral dissertation. The main research works are stated as follows:1. For a class of uncertain single input and single output (SISO) strict-feedback nonlin-ear system, an adaptive fuzzy tracking control via output feedback is proposed. The dynamic feedback strategy begins with an input-driven filter. By utilizing fuzzy logic systems to approx-imate unknown and desired control input signals, an output feedback adaptive fuzzy tracking controller is designed via backstepping design method. By choosing an appropriate Lyapunov function, the proposed control method can guarantee all the signals in the close-loop system to be semi-globally uniformly ultimately bounded, and the tracking error converges to a small neighborhood of the origin. Simulation results are presented to demonstrate the effectiveness of the proposed methods.2. For a class of uncertain SISO nonlinear systems with unknown time-delay functions, an adaptive fuzzy output-feedback control method is proposed. The dynamic feedback strategy begins with an input-driven filter. In the backstepping design process, dynamic surface control technique is used to avoid repeated differentiate certain nonlinear functions. Therefore, the computation burden is greatly reduced and the designed controller is simplified. It can be proved that the proposed control method can guarantee all the signals in the close-loop system to be semi-globally uniformly ultimately bounded, and the tracking error converges to a small neighborhood of the origin. Simulation results are presented to demonstrate the effectiveness of the proposed methods.3. For a class of SISO strict-feedback stochastic nonlinear systems with unknown time-delays, an observer-based adaptive neural network control method is proposed. First, a state observer is designed, then an adaptive output-feedback controller can be obtained via backstep-ping design method. In the backstepping design process, by introducing a quartic Lyapunov- Krasovskii functional, time-delay terms in the system can be compensated, and the designed controller can guarantee all the signals in the closed-loop system to be semi-globally uniformly ultimately bounded in mean square. In addition, dynamic surface control technique is used to avoid the explosion of computational complexity in the backstepping design process. At last, simulation results are presented to demonstrate the effectiveness of the proposed methods.4. For a class of strict-feedback large-scale stochastic nonlinear systems, an observer-based adaptive neural network decentralized control method is proposed. As large-scale non-linear system is complex and high in dimension, decentralized control method is used to re-duce the amount of information to be dealt with. By combing backstepping design method and neural network, a direct adaptive neural network controller is designed, and the number of adaptive parameters needed to adjust online is greatly reduced. In addition, dynamic surface control technique is used to avoid the explosion of computational complexity in the backstep-ping design process. At last, simulation results are presented to demonstrate the effectiveness of the proposed methods.5. For a dynamic system of robotic finger, an adaptive fuzzy tracking control method is proposed. Based on researches on fuzzy logic systems and adaptive control method, an adap-tive fuzzy decentralized controller is designed. The number of adaptive parameters obtained in the controller design process is greatly reduced, therefore the designed controller is easy to apply in practice. Finally, simulations are exerted on Puma560robot manipulator to show the effectiveness of the developed control method.