Study On Some Control Problems For A Class Of Nonlinear Systems

In this thesis, multiviable affine nonlinear systems are studied. The systems are classified into two categories, minimum phase and nonminimum phase, based on differential geometry feedback linearization, and then we study control and adaptive fault tolerant control prolems of the two class of systems. A set of control and adaptive fault tolerant control methods are proposed for multivariable affine nonlinear systems using feedback linearization, model prodictive control, adaptive fault tolerant control, vanishing perturbation theory, etc.Firstly, the robust adaptive contol problem for a class of multi-input multi-output (MIMO) nonlinear minimum phase systems with uncertainties and disturbaces presented by input-output models are studied. An adaptive tracking controller is presented by using state feedback and output feedback based on neural networks modeling for sytems'unknown nonlinearities. And then, a robust compensation item is proposed by Lyapunov redesign method, the robust tracking performance is guaranteed.Secondly, an exponential stabilization scheme is presented using feedback linearization and model predictive control for affine MIMO nonminimum phase systems. The inputs-outputs of the systems are decoupled by feedback linearization, and the external dynamics of the systems are stabilized by high-gain state feedback, then model predictive control is used to stabilize internal dynamics. The proposed controller can guarantee that the closed-loop systems are exponential stable. Using the result above, an exponential tracking controller is designed, and a compensation item is proposed by neighboring extremal theory, the robustness of the systems are improved.Further, adaptive fault tolerant control methods are studied for affine nonlinear systems with actuator failures.A model reference adaptive fault tolerant control scheme is designed for multi-input single-output (MISO) affine nonlinear minimum phase systems with actuators lock-in-space or/and loss of effectiveness failures, the asymptotic output tracking control prolem of the systems is solved; and then, for the MISO affine nonlinear minimum phase systems with actuators lock-in-space or/and variant actuator failures, using neural networks modeling for variant actuator failures online combination with model reference adaptive fault tolerant control scheme, the bounded tracking control problem of the systems is solved.A neural adaptive control scheme is proposed for MIMO affine minimum phase systems with actuator lock-in-space or/and variant actuator failures. The actuators are classified with the characters of the system, and then model reference adaptive control scheme is proposed using neural networks modeling for time varying actuator failures online, the bounded tracking control prolem of the systems is solved.Finally, an exponential stabilization scheme is proposed for MISO affine nonminimum phase systems by using vanishing perturbation theory and approximately feedback linearization, and the domain of attraction is presented. An adaptive fault tolerant control scheme is presented using the result above, adaptive stabilization prolem for the systems with actuators lock-in-space or/and loss of effectiveness failures is solved.Digital simulations are proposed by the model of manipulator, aircraft, etc, simulation results demonstrate the effectiveness of the proposed methods.