Research On The Decoupling Control For Multivariable Nonlinear Systems Using Inverse Method

In reality, many industrial processes are complex multivariable non-linear systems. However, the traditional controlling and theories are mainly based on the linear systems. So, it will have significant theoretical value and practical meaning to develop decoupling control schemes for the multivariable non-linear systems, which not only can provide excellent decoupling and controlling performance but also can be conveniently applied to practical application.In this thesis, the research and controlling object is the typical multivariable non-linear system and we use the inverse system to decoupling control the object. As follows are the concrete research contents:(1) This thesis presents and proves the necessary and sufficient condition for the existence of the inverse model in a multivariable non-linear system. We find the analytical form of the inverse model based on two different descriptions. We not only discuss the decoupling ability of the pseudo-linear system in this thesis, but also get the conclusions for decoupling stabilization of the pseudo-linear system;(2) On the basis of the theory of inverse system and the analytical form of the generalized inverse model, this thesis research the implementation method and the structure of the neural network (NN) generalized inverse decoupling control system for the typical multivariable non-linear system. We make simulation experiment under the MATLAB environment;, and the result indicates the great decoupling ability and certain robustness of this decoupling control system;(3) In order to the problems existing in the practical applications, we propose a new decoupling method which combines the IMC and the inverse system in this thesis. We not only analyze the closed loop steady state performance of the pseudo-linear subsystem, but also study the conclusions for robust stability of the IMC system;(4) This thesis preliminarily studies the robust stability of the two-degree freedom and2-port IMC, then presents and proves the conclusions for the robust stable of the closed loop system. Finally, we make simulation experiment under MATLAB/SIMULINK environment; the result indicates that in comparison to other structures with inverse system, this method has simpler controller, excellent control effect and stronger robustness.