The Analysis And Design Of Multirate Predictive Control Systems

This thesis discusses several researches on multi-rate predictive control contenting of two parts. The first part is the algorithm design, which consists of dual-rate predictive control based on the pole placement, dual-rate min-max robust generalized predictive control and dual-rate predictive control for time-delay systems; the second part is the performance analysis of multi-rate predictive control systems.The main results are: (1) The IMC structure of the dual-rate predictive control system is derived, and its robust stability and zero steady state error characteristics are analyzed. Simulation examples explain the dual-rate predictive control has more superiority than the single-rate predictive control. (2) An algorithm of dual-rate predictive control based on the pole placement is proposed. A class of system that the control is applied at a faster rate than the available measurements of the plant output signal is chosen to illustrate the technique. A state feedback gain matrix is designed, which can not only assign closed loop poles at desiring position but also ensure the system robust stability, though there are some uncertain parameters in the system. This result precedes the former research work. (3) Dual-rate min-max robust generalized predictive control algorithm is proposed, which adds a constraint output horizon to improve the system properties. A controller with strong robustness is derived, which is robust to disturbances by minimizing the cost function while disturbances maximize it. Some robust stability conditions are derived. (4) A class of dual-rate input-output predictive model containing time delays is established. A new dual-rate predictive control algorithm for the time-delay system is presented, which makes up the shortcoming of multi-rate weighted minimum-variance control for time-delays system, and has solved the influence of time delays on the multi-rate system. Furthermore, the robust stability of the closed-loop system is analyzed using linear matrix inequalities (LMI) method. (5) Computer simulations demonstrate the usefulness of the proposed algorithms in this paper.