Research On Predictive Control Of A Class Of Hybrid Systems With Constrained

With the further research and widely used of control science and computer technology, a multitude of advanced systems have emerged in the areas such as traffic, manufacture, communication etc. Many of them have been accomplished by innovation in techniques, such as the applications of sophisticated embedded control systems. Among those systems, there are not only physical devices such as mechanism, fluid equipment etc, which evolve following physics law, but also logic units such as computer, PLC etc, which operate according to DEDS (Discrete Event Dynamic System) rules. Thus there exist two kinds of different variable and mode respectively, they are interaction each other. Usually, we call it hybrid systems. On the other hand, the system variables have been limited to a bound, generally. So, hybrid systems with constrained is a more appropriated name. It can describe and process a class of practical systems and has an important research value.In constrained hybrid systems, a state space has been partitioned into several convex sub-space called cell, which are governed by different differential (difference) equations, every of them can be considered as a separated constrained linear system. The system switched among them according to logic rules. Therefore, constrained linear system is a foundation for the study of the constrained hybrid systems and plays an important role in mastering the principle of the latter. The paper begins with the study of the control of constrained linear systems and consists of following four aspects:(1) Infinite time optimal control of constrained linear system. By means of terminal inequality constraint set which is a maximal output admissible set, the problem has been transform into a finite time optimal control of constrained linear systems and an infinite time constrained LQR problem. Then the former has been worked out though the Mp-QP (Multi-parameter Quadratic Programming). Besides, the optimality and stability can be obtained simultaneously.(2) Predictive control of constrained linear systems. RHC (Receding Horizon Control) can be achieved by repeated optimal control on current state and convergence property has been proved.(3) Modeling of constrained hybrid systems. Piece-wise linear system with logic rules has been separated into two relative independent systems——CVDS(Continue Variable Dynamic System) and DEDS, meanwhile, two interfaces have been designed to interconnect between them, one is event trigger, the other is mode selector. Then, propositional logic has been introduced to describe the DEDS,auxiliary variable has been set up to modeling the interface.(4) Predictive control of constrained hybrid systems. Optimal control of constrained hybrid systems has been introduced firstly and transformed into MIQP(Mixed-Integer Quadratic Programming)so as to obtain the control input series. On the basis of which, predictive control strategy has been worked out by means of receding optimization.At the end of thesis, the research work is summarized, and the prospects of the research field are outlined.