电子社现代控制理论基础（英文版）（第2版）mooc课件6-5.pptx

Consider an uncontrollable LTI system;Theorem： If an LTI system (A,B,C,D) is not controllable, then the necessary and sufficient condition for stabilizing the system by state feedback is: the uncontrollable states are asymptotically stable.;It can be seen that;The characteristic polynomial can be written as;We can see that the eigenvalues of are not consequently affected by the state feedback. The characteristic polynomial of the closed-loop system is;Consider a system;Using state feedback , where , we can assign desired poles at . Therefore the desired characteristic polynomial is ;Thus the state feedback gain can stabilize the original system. The closed-loop poles are located at and -5. We can verify the conclusion as;;Decoupling is an important content in studying most of MIMO systems. The design objective is seek appropriate control law such that the MIMO systems can be treated as several SISO systems. Generally, the decoupling can be done by cascaded or feed forward compensators or state feedback.; 2.Decoupling by state feedback ;Let the rows of the output matrix be written by . The necessary and sufficient condition for being able to decouple state-space representation by state feedback is that there exists a nonsingular matrix:;When , takes . When ，take the minimum .;The closed-loop transfer function can be further derived as;Consider a system as;The simulation diagram is shown as :;If we make , then;Furthermore, we have ;The transfer function is;The answer to this problem is yes, provided dynamic feedback compensators (observers for example) of sufficiently high dimension are allowed.