Research And Application Of Predictive Functional Control

Model predictive control (MPC) is one of the most popular and successful advanced control strategies, which has the advantages of good performance and strong robustness. Predictive functional control (PFC) is the third generation MPC, and has the main characteristics of traditional MPC. Besides,the manipulated variables of PFC are structured, which can be represented as the combination of some pre-determined basis functions. Thus the manipulated variables are more regular and the online amount of computation is reduced. In this paper ,both theory and application of PFC are studied as follows:(1) A survey on the development,status quo and industrial application of MPC and PFC is presented, and the general principles of MPC and PFC are introduced.(2) The PFC algorithms based on first-order plus time-delay model and second-order model are introduced respectively. A single value PFC algorithm based on Controlled AutoRegressive Moving Average (CARMA) model is proposed, which reduces the online computation. The closed-loop performances of these PFC algorithms are analyzed based on the Internal Model Control(IMC) structure . Then the basic principles of the parameters selection of PFC are given and the performance is shown by comparation in the simulation.(3) The on-line computation of multivariable predictive control algorithm sharply increases when the dimensions of system and the prediction horizon increase, thus the real-time performance of the algorithm cannot be guaranteed. A multivariable predictive functional control algorithm based on CARMA model is proposed in this paper. The basis functions are simplified in the prediction horizon, thus the number of variables to be optimized is reduced; the predictive outputs of the multivariable model are obtained through recursive operation,which can avoid solving Diophantine equations and separate the coefficients of the variables to be optimized ,thus the online computation is reduced. A simplified method of dealing with intput constraint is given. The closed-loop performance and freedom degrees of multivariable PFC are analyzed. Finally, the simulation results demonstrate the feasibility and effectiveness of the algorithm.(4) An advanced control scheme for the delayed coke furnace is designed due to the current control problems , and an advanced control software is developed for the system. The sofware implements the predictive functional control of coke furnace, and provides very good operation interface for users. The software also have a strong fault-tolerance and generalization which can be used in other control system with just a little modification.The last part is the summary of this thesis and the future research directions are discussed.