Research On The Application Of Internal Model Control In Multivariable System

Control over multi-input multi-output (MIMO) system is always a difficult task in the field of control. In industrial manufacturing process, MIMO process with the features of time-delay and coupling are ubiquitous. As to the time-delay multivariable system, diagonal decoupling matrix acquired by traditional methods has differentiation element, which is physically impossible, thus making the system more difficult to control. In addition, multivariable control system not only bear multi-coupling and multi-time-delay, but also is complicated in its structure, with input and output unequal, while the transfer function is a singular matrix. Traditional multivariable internal model control is based on the inversion of nonsingular object, and therefore it is difficult to resolve this problem.In this paper, a controller is designed by using internal model control strategy. Attributes of multivariable system and internal model control methods of multivariable square system and of multivariable non-square system are studied and analyzed. Controllers designed in the methods introduced in this paper can overcome difficulties of the traditional PID in realizing the optimality of control effects, can realize decoupling control over multivariable system, the system with multi-time-delay and multi-decoupling in particular. Compared to old control methods, the new one is better at resistance against process disruption, tracking and maintaining stability.This paper touches on basic principles, structures, natures and robustness of multivariable internal model control as well as typical design methods of multivariable internal model controller. Focus is given to the study of the internal model of category I multivariable square system with multi-time-delay and multi-decoupling. To effectively avoid decoupling errors resulting from model mismatches, the proposed designing method in this paper handles decoupling and control simultaneously. The method does not only bring sound control effects but also displays relatively strong robustness.The specific characters of the structure of multivariable non-square system are analyzed in this paper and major control methods of multivariable non-square system are studied. Two methods are studied. One is to make non-square system square, namely matching through relative gain matrix and the other is direct control method, represented by decoupling non-square system. Besides, this paper solves the problem of accurately control thin system through selecting control methods. Thin system is rarely seen in chemical production process.Based on singular value decomposition, the paper for the first time puts forward a new method for internal model control to replace non-square system where the amounts of input variables and output ones are not equal. Non-square system frequently takes place in complicated industrial production process. Based on SVD array theory, a non-diagonal filter is designed, which enables the control system not only to have the advantages of high-dimensional decoupling capacity and fast response ability but also to be strongly robust.