Tracking control of nonlinear systems based on online parameter and gain adaptation

Tracking control is an important area in control theory. It appears in many applications such as robot manipulators, automobiles, airplanes, etc. As virtually all physical systems are nonlinear in nature, researches on complex control systems with capability of handing the nonlinearities are in great demand. During the past decades, a large variety of approaches has been proposed in this field. For tracking control of a dynamic system with uncertainties, it would be best to know the precise model. However, even if the structure information can be obtained, parameters of the structure are usually unknown. Adaptive control has been developed by many researchers to deal with the structured system with unknown parameters. With parameters on-line updating, the system can be asymptotically stable in theory. Nevertheless, it often gives poor performance in the practical application because the difference between discrete-time and continuous-time form cannot guarantee the system stability. In this thesis, a novel control scheme based on real-time parameter and gain adaptation is proposed to realize the tracking control of nonlinear systems. The control system consists of two different controllers. The feedforward controller is developed to handle the deterministic nonlinearities and uncertainties. The parameters of the system are on-line estimated effectively and converge to their actual values as the process going on. The feedback controller is developed to reduce the unmodeled nonlinearities and disturbance. The gain of the controller is adaptive based on error pattern analysis along the sliding surface, hence improving the control performance evidently. To verify the effectiveness of the proposed control scheme, simulation and experiment are conducted and the results indicate its usefulness in tracking control system