Robust Predictive Control For Time-delay System And Its Application Research In Thickness Control System Of Rolling Mill

With the increasing production requirements and increasingly complex workingconditions of modern industrial systems, there are often time-delays and various uncertainfactors in industrial control systems, and the thickness control system of rolling millactuated by hydraulic servo system can be approximated as a linear uncertain time-delaysystem. In addition, in the actual control system, the existence of time-delay anduncertainty usually affect performance of the control system seriously, and may even leadto instability for the system, which cause a great deal of difficult for the design of thesystem. In recent years, the rapid development of robust predictive control theory, whichcombines the receding horizon principle of predictive control with the uncertaintiesprocessing method of robust control, provides a possibility to design a controller withstrong robustness for the time-delay system with uncertainties.Firstly, this paper introduce the basic theory of linear matrix inequality (LMI) simply,and the basic principles and qualitative synthesis of predictive control are elaborated,which laid a theoretical foundation for further analysis and design of the system.Secondly, a robust predictive control algorithm based on memory state feedback isproposed for a class of uncertain discrete time system with input constraint and multiplestate delays. By introducing Lyapunov-Krasovskii function and using linear matrixinequality method, the algorithm can convert the infinite time domain optimizationproblems into linear programming problems. Sufficient conditions of the existence andexplicit expression for the memory state feedback controller are obtained, and the controlinput can satisfy constraint as well. Asymptotical stability of the closed-loop system andminimum of robust performance index can be guaranteed simultaneously.Finally, an H-infinity robust predictive control method based on memory statefeedback is proposed for the thickness control system of rolling mill, which containsparametric uncertainty and external disturbance. The method combines the robustH-infinity control theory with the receding horizon principle of predictive control, and thesufficient conditions for the existence of the memory state feedback controller are given by using Lyapunov stability theory and linear matrix inequality technique. Furthermore,the design methods of the controller are proposed by solving a linear matrix inequalityconstrained convex optimization problem. The obtained controller can not only make thethickness closed-loop control system of rolling mill asymptotical stability, but also achievethe given H-infinity performance.