LMI Minicx Problem (Robust Control)

Question

yan zong 2019-10-22

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/486887-lmi-minicx-problem-robust-control

评论： REVATHI 2024-10-22

采纳的回答： yan zong

在 MATLAB Online 中打开

This Theorem is from one paper published in IEEE Trans Automatic Control, the details of theorem are as follows:

---------------------------------------------------------------------------------------------------------------------------

Then, I using the following configurations to test this theorem

alpha = -0.07, and beta is -0.28, delta is equal to 0.5.

--------------------------------------------------------------------------------------------------------------------------

If I want to find the miniman vlaue of gamma, i have the following LMI equation

Minimise gamma subject to (10)

-------------------------------------------------------------------------------------------------------------------------

Finally, I develop my own code, please see the following. However, it cannot work well, is there anyone can provide me some advice? Thanks a lot.

%% system definitation
delta = 0.5;
A = [delta 0.8 -0.4; -0.5 0.4 0.5; 1.2 1.1 0.8];
B = [0 1; 2 -1; 0 1.3];
E = [0.1; 0.4; 0.1];
C1 = [-1 0 2];
D = [0 0];
F = 0.3; 
C2 = [-1 1.2 1; 0 -3 1];
H = [0.1; 0.4];
alpha = -0.07;
beta = -0.28;
A_bar = [A E; C1 F];
B_bar = [B; D];
% K_bar = K;
C_bar = [C2 H];
%% define the LMI system
setlmis([]);
%% Defining Variables:
gamma = lmivar(1, [1 0]); % gamma(1,1)
[P, n, sP] = lmivar(2, [2 2]); % P(2,2)
[G, n, sG] = lmivar(2, [4 4]); % G(4,4)
[V, n, sV] = lmivar(2, [2 2]); % V(2,2)
[U, n, sU] = lmivar(2, [2 2]); % U(2,2)
[J, n, sJ] = lmivar(2, [4 4]); % J(4,4)
[Xi_11, n, sXi_11] = lmivar(3, [sP, zeros(2, 2); zeros(2, 2), -gamma*gamma*eye(2, 2)]);
[Xi_22_bar, n, sXi_22_bar] = lmivar(3, [sP, zeros(2, 2); zeros(2, 2), eye(2, 2)]);
[Xi_22, n, sXi_22] = lmivar(3, [-alpha*(sG)-alpha*((sG)')+alpha*alpha*(sXi_22_bar)]);
%% Defining LMIs term contents:
% DEFINITION 1-st row
lmiterm([1 1 1 Xi_11], 1, 1); % #1 LMI, the (1, 1) block
% DEFINITION 2-nd row
lmiterm([1 2 1 G], 1, A_bar); % #1 LMI, the (2, 1) block
lmiterm([1 2 1 V], B_bar, C_bar); % #1 LMI, the (2, 1) block
lmiterm([1 2 2 Xi_22], 1, 1); % #1 LMI, the (2, 2) block
lmiterm([1 2 2 J], 1, 1); % #1 LMI, the (2, 2) block
% DEFINITION 3-rd row
lmiterm([1 3 1 V], ((B_bar')*B_bar), C_bar); % #1 LMI, the (3, 1) block
lmiterm([1 3 2 0], 0); % #1 LMI, the (3, 2) block
lmiterm([1 3 3 U], beta*((B_bar')*B_bar), -1, 's'); % #1 LMI, the (3, 3) block
% DEFINITION 4-th row
lmiterm([1 4 1 0], 0); % #1 LMI, the (4, 1) block
lmiterm([1 4 2 0], 0); % #1 LMI, the (4, 2) block
lmiterm([1 4 3 G], 1, B_bar); % #1 LMI, the (4, 3) block
lmiterm([1 4 3 U], B_bar, -1); % #1 LMI, the (4, 3) block
lmiterm([1 4 4 J], (1/(beta*beta)), -1); % #1 LMI, the (4, 4) block
lmisys = getlmis;
%% check the feasibility
% [tmin, xfeas] = feasp(lmisys);
% V = dec2mat(lmisys, xfeas, V);  
% U = dec2mat(lmisys, xfeas, U);
%%  Defining vector "c" for C'x in mincx
Num = decnbr(lmisys);
c = zeros(Num,1);
c(Num)=1;
%% Solving LMIs:
[copt,xopt] = mincx(lmisys,c);
%% Finding Feedback gain and u:
% display(gopt)
% K = inv(U)*V