Study Of Improve Generalized Predictive Control Algorithm

Along with the rapid development of industrial production, technological advances and computer technology, a new type of computer control algorithms called Predictive Control has also achieved rapid development. Those originally typical Predictive Control Algorithms such as Model Predictive Heuristic Control (MPHC), Model Algorithmic Control (MAC) and Dynamic Matrix Control (DMC) are proposed respectively by Richalet, Mehra and Cutler. There are three aspects of the main features for Predictive Control:predictive model, receding optimization and feedback compensation. These advantages can eliminate the effect of the controlled object owing to the factors like external environment, parameters, object modeling errors and structures. This kind of algorithm is not only with good robust characteristics but also with great control. Generalized Predictive Control (GPC), as a type of Predictive Control, has been developed with the Adaptive Control Method, which is proposed by Clarke etc. GPC is a type of Predictive Control Algorithm based on Parametric Model, which not only keeps the advantages of the typical predictive control algorithm such as MAC and DMC based on the non-parametric model, but also combines identification and self-correcting mechanism. Consequently, it shows characteristics of great robustness. For these reasons, GPC has got more and more attention for study.Firstly, the thesis gave the introduction of the basic algorithms for GPC, and based on the example of SISO (Single Input Single Output) Systems, which analyzed the impact of the changed controller’s parameters to system performance by using simulation. Secondly, the thesis aimed at traditional GPC which use Recursive Algorithm to avoid Online Solving Diophantine Equation, but with the problems of large computation and numerical morbidity. Thus, the Stair-like General Predictive Control had been proposed. And basing on the disadvantages of long line occupancy and large computation of Display Algorithm of GPC, another type of Improved Generalized Predictive Control Algorithm which can also be called Implicit Self-turning Algorithm is presented. The equivalence of control rate for DMC and GPC are applied in this algorithm. And then, the research combined the Stair-like General Predictive Control and Implicit Self-turning General Predictive Control together, which had the advantages of both two algorithms. And the combined algorithm was applied into SISO System by using Simulation Research to confirm the feasibility and validity.Finally, this thesis extended this combined algorithm to multivariate control system. During this process, the coupling control problem in multivariable system for decupling control problem is designed. And, verified the feasibility and validity of improved algorithm in multivariable system by making use of MATLAB. Then, simulation research was carried out aiming at how to adjust the parameters in the MIMO system.