| With the development of the industry, the multivariable system in which lots of parameters interact appears in industrial process of control. Therefore, the design of decoupling for the multivariable system is most important in the process of control.For the interacted system, two methods are given in order to judge the coupling degree of the multivariable system, such as relative gain method and Inverse Nyquist Array method. For the strong coupling system which needs to be decoupled, this paper introduces the principle and conditions of complex frequency domain decoupling method and time domain decoupling method .And the corresponding examples are simulated and analyzed. For feed-forward compensator decoupling method, a judgment method is proposed. By using the method, we can know whether the feed-forward compensator decoupling method is appropriate to the multivariable linear system or not. The method is based on minimal design problems. The minimal design problem is applied to the decoupling problem. And the conditions of the feed-forward compensator decoupling are obtained from the method. For the complex high order system, diagonal dominance is introduced. Thus, the high order system achieves approximate decoupling.The robustness, anti-interference ability and ranges of predictive control parameters to ensure the decoupling control performance of the multivariable linear system are analyzed when the dynamic matrix predictive controller is designed for the system. Feed-forward compensation decoupling method, state feedback decoupling method and static decoupling method are combined with dynamic matrix predictive controller. Simulations of decoupled multivariable linear system are given. Analysis of the robustness, anti-interference ability and ranges of predictive control parameters to ensure the decoupling control performance of the decoupled system are also given. Open-loop predictive decoupling has better robustness from analyzing the simulation experiment. |
