Stable Continuous-time Generalized Predictive Control

This thesis discusses the closed-loop stability of the continuous-time generalized predictive control, which consists of two parts. The first part studies the asymptotically stability of the closed system, the second part investigates the closed-loop stability of the continuous-time generalized predictive control with constraints, model uncertainty and time delay, respectively.The main results are: (1) The new methods to guarantee the closed-loop stability of the continuous-time generalized predictive control are given. The cost function with infinite horizon is imported. By the new end-point equation constraints, the optimal problem which has a cost function with infinit horizon is transferred to an optimal problem with finite horizon. The condition which can guarantee closed-loop system of continuous-time generalized predictive control to be stable is given according to the proof of the monotonicity of the receding-horizon cost. (2) For the continuous-time generalized predictive control with constraints on control input, on system output and derivative of system output and on any intermediate variables, the method of the cost function with infinite horizon can guanrantee the closed-loop stability of the constrained continuous-time generalized predictive control. (3) The continuous-time generalized predictive control with model uncertainty is investigated. The system parameters are estimated by using the continuous least-squares estimator, and the cost function with infinite horizon is imported based on the estimated system model. Then the method of guaranteeing the closed-loop stability of the continuous-time generalized predictive control with model uncertainty is given. (4) The continuous-time generalized predictive control with content time-delay is investigated. The dealy-dependent control law can be obtained by the basic algorithom of continuous-time generalized predictive control. The cost function with infinite horizon, which is connected with time delay, is used. Then the optimal control law can guarantee the closed-loop stability of the continuous-time generalized predictive control with time delay.