A Study On The Theories And Applications Of Model Predictive Control

Model predictive control is a kind of advanced control technology, which is based on themodel of plants. The model predictive control was straightly engendered and sophisticated fromthe applications of industrial process. Model predictive control requires low accuracy of modelprecision, but it can provide high performance. Model predictive control is a quite effectiveoptimal control algorithm, which made up the deficiency of modern control theory in complexplants. In the recent years, with the development of many kinds of the control theory, such asintelligent control, PID control, wavelet theory, robust control and hybrid systems control, thetheory of model predictive control has also obtained rapid developments.Based on the study status of model predictive control and the new requirements from thepractice industrial processes for the theory of model predictive control, some desideratedproblems for the theory of model predictive control, corresponding results are given. The maincontents are as follows:1. For the polytopic uncertainty system, we present a new technique for the synthesis of arobust model predictive control law, using linear matrix inequalities. Firstly, we formulatedthe robust unconstrained MPC problem. We then extend the formulation to incorporate inputand output constrains.2. More common, paradigm for robust control consists of a system with uncertainties appearingin the feedback loop. The structured feedback uncertainty models a number of factors, suchas nonlinearities, dynamics or parameters, that are unknown or unmodeled. For the structureduncertainty system, we present a robust state-feedback MPC controller based on LMIs.3. The state-feedback control law has so many extrusive virtues, but in a number of practiceproblems the system states are not easy to obtain directly. This causes difficulties in physicalimplement. Therefore how to conform a state estimator to replace the accurate states infeedback control is a problem. This paper we design a state-feedback estimator for polytopicuncertainty system. The error uncertainty caused from model uncertainty was calculated asdisturbance instead of an external signal, which decreased the mismatch of state estimatorand improved characteristics of feedback control law.4. The feasibility of receding horizon control law that we obtain and robust asymptotic stabilityfor the closed-loop system was proved.5. Using the GPC parallel predictor to identify the farthest remote output parameter directly.And ascertain optimal control law by proves the equivalence property of GPC and DMC.Instead of only using the first term of input increment, an input weighted control law,possessing rectifier filter function, was introduced into the algorithm, which can make thebest use of the subservience information in multistep prediction. Thus we present animproved implicit GPC self-correction controller algorithm, which preserving hardrobustness and have well control performance in stable minimum-phase, non-minimumphase system as well time delay system.