Research And Application Of Multivariable Predictive Functional Control

Model predictive control (MPC) refers to a class of computer control algorithms. At eachcontrol interval an MPC algorithm predicts the future response of a process, and attemptsto optimize future process behavior by computing a sequence of future manipulatedvariables. Because this kind of control algorithm has taken into consideration of theindustrial requirements, the control system under MPC has good performance and strongrobustness. Predictive functional control (PFC) is a branch of the MPC family. The maindistinguishing feature of the PFC algorithm over other MPC algorithms is the constructionof the manipulated variables. Usually, the manipulated variables can be represented as asum of pre-determined basis functions, which achieves computational simplicity and goodperformance of tracking set-point without steady-state error. Some problems of predictivefunctional control are researched in this thesis, and the main research works are as follows:(1) An overview of MPC and PFC technology is given.(2) A multivariable predictive functional control algorithm based on a two-inputs/two-outputs system with the transfer function model is presented in this section. A simpleand explicit solution of manipulated variables of the control system can be obtained byoptimizing the objective function. The stability and robustness analysis of this kind ofmultivariable predictive functional control algorithm is given. Finally, simulations andexperiment of the system applying this control algorithm are provided, showing that thepresented algorithm is feasible.(3) A multivariable predictive functional control algorithm, on the basis of a three-inputs/three-outputs system with the transfer function model is presented, which has beenpromoted into the n-inputs/m-outputs system. The explicit solution of manipulatedvariables is obtained by using the idea of Smith predictor. In the end, some simulationexamples are presented by using Shell Oil's heavy-oil fractionator model. The result ofcomparing with single-variable predictive functional control is shown in the simulation,which displays that this multivariable predictive functional control algorithm has goodcontrol performance.(4) Based on the mechanism of plastic melting, the energy conservation mathematicalmodel is proposed. The parameters of this model are obtained by identification online,which ensures the robustness and disturbance rejection. An advanced control system has been established which based on the model and the predictive functional control algorithm.The good result of experiment is shown in the last part, approving the validity of thecontroller.The last part is summary and perspective.