Model Predictive Control For Hybrid Systems Based On Mixed Logic Dynamical Model

In many industrial process control areas, the systems to be controlled is not only constituted by continous parts, but also constituted by parts described by logic, such as for instance on/off switchs or valves, gears selectors.This two parts mixed and interact with each other. It's difficult to describe these systems with differential or difference equations in traditional control theory framework. So it's urgent to build a new theory framework, which is capable of dealing with the continous and logic parts and their interaction.Hybrid system is such a system whose behavior is modeled as the interaction between continuous systems and discrete-event systems.The research on hybrid systems provides new ideas and methods for modeling and controlling large complex systems. Based on hybrid system theory, industrial process control can take full advantage of differernt kinds of knowledge, such as artificial intelligence and computational intelligence, operational research and propositional logic.On the basis of summarizing and reviewing previous research work on hybrid system, this dissertation makes some further research work and explorations in hybrid system modeled by mixed logical dynamical methods and its model predictive control based on the model. Mixed logical dynamical model can be looked at as an extension of linear continous dynamic model to hybrid systems.It can model the continous and discrete parts and their interactions in an integrated model framework.And it also can integrate the logical rules, heuristics and operating constraints into the model. The main research work in the thesis is as follows:(1) The background and the concept of the hybrid systems, the importance and contents of research of hybrid systems are explained. Differnent typical widely used modeling approaches for hybrid systems and their characteristics are summarized. The stability analysis and the optimal control methods of hybrid systems are reviewed specially. Typical examples of the applications of hybrid systems in engineering practice are also illustrated.(2) Some basic facts for transforming propositional logic problems into equivalent mixed-integer linear inequalities are reviewed, which are the main tools for acquiring the mixed logical dynamical model of the hybrid systems. And the main tools for modeling Principle modeling approach are deeply analyzed and acquired based on Mixed Logic Dynamic (MLD). The structure of the hybrid control system and its mixed logical dynamical model are studied. And the general description expression of the mixed logical dynamical system is discussed. Several typical systems modeled by the mixed logical dynamical methods are provided as examples.(3) The key ideas and main characters of model predictive control(MPC) are reviewed. The stability definitions of the mixed logical dynamical systems and related issues are discussed.And the optimal control of mixed logical dynamical systems,which results in a mixed-integer quadratic programming (MIQP) problem, is studied.Using the mixed logical dynamical model as the predictive model, the predictive control for hybrid system with the operating and state constraints is researched based on the minimization of the quardratic performance index. And also the stability of the closed loop system is proved under the control law.Despite the fact that very effective methods exist to compute the optimal solution of the MIQP problem, the compution burden of solving the MIQP problem online is still large.In princible, this might limit the scope of application of the control method to very slow system.So we study the predictive control of the MLD system based on the linear performance index,which leads to a mixed-integer linear programming(MILP) Problem. Simulation results show the rationality and validity of this two predictive control methods.(4) The threetank hybrid system is a typical hybrid system with constraints and is meaningful for the research on the fault dectection and control algorithms of the industial process control.The MLD model of the system is constructed.And by using the MLD model as the predictive model, two predictive control algorithms, respectively based on the quardratic and linear performance index , are designed to keep each tank remain the desired level.The simulation experiment is done in the Matlab envoriment.By analysing the simulation result, the conclusion is reached that the two predictive control laws can control the threetank system well and are robust and anti-fault.At the end of the thesis, the research work of this thesis is summarized, and prospects for further research work are outlined.