Control Avanzado

Edgar Armando Chulines Domı́nguez

April 10, 2017

I. Home task 1: Using Schur’s complement, find α and β for which S > 0

10 3 α 1

3 β 0 0
S=
α

0 0 1

1 0 1 2

Solution:

By partioning S matrix, we obtain

     
10 3 α 1
 3 β  0 0  

S= ,
 α 0 0 1 
1 0 1 2
       
10 3 10 3 α 0 0 1
where S11 = , S12 = , S21 = & S22 =
3 β 3 β 1 0 1 2

Now by Schur’s complement: S11 − S12 ∗ S22 ∗ S21 > 0, we obtain

−1

     −1  
10 3 α 0 0 1 α 0
− ∗ ∗ > 0 =⇒
 3 β   1 0  1 2   1 0
10 3 α 0 −2 1 α 0
− ∗ ∗ > 0 =⇒
3 β  1 0 1 0  1 0
10 3 −2 ∗ α + 1 α α 0
− ∗ > 0 =⇒
3 β 0 0 1 0
   2   2 
10 3 −2 ∗ α + 2 ∗ α 0 2 ∗ α − 2 ∗ α + 10 3
− = >0
3 β 0 0 3 β

We apply the Schur’s compliment one more time to the last matrix,

 2 
2 ∗ α − 2 ∗ α + 10 3
>0
3 β
2
S11 > 0 =⇒ 2 ∗ α − 2 ∗ α + 10 > 0 & S22 > 0 =⇒ β > 0

1
 then the intervals are
2
For a:−∞ < α < ∞ since 2 ∗ α always positive and greater than −2 ∗ α,
2
then the left side of the inequality 2 ∗ α − 2 ∗ α + 10 > 0 will be at least 10.
For b: β > 0.

II. Home task 2:

1 β
Q=
, F = (1, 1)
β −1

β−? xT Qx < 0, for ∀x : f x = x1 + x2 = 0

X T QX < 0
Q − µF T F < 0
   
1 β 1 1
−µ <0
β −1 1 1
   
1−µ β−µ µ−1 µ−β
< 0 ⇐⇒ <0
β − µ −1 − µ µ−β µ+1

By sylvester criterio: µ − 1 > 0, µ > 1:

(µ − 1)(µ + 1) − (µ − β)(µ − β) > 0

µ2 − 1 − µ2 + 2βµ − β 2 > 0
2βµ − β 2 − 1 > 0
1 + β2
µ> >1
2β
1 + β 2 > 2β
β 2 − 2β + 1 > 0
(β − 1)2 > 0
β>1

2
III. Home task 3: Using s-procedure to find τ1 ≤ 0 such that S(x) = F (x) −
τ1 G1 (x) < 0 where f (x) = xT F x < 0 whith

1 −1 0

F =
−1 −2 0

0 0 −3
For all x satisfying

2 −1 0

G1 (x) = xT F1 x ≤ 0 whith F1 =
−1 1 0

0 0 1
   
1 −1 0 2 −1 0
−1 −2 0  − τ −1 1 0 >0
0 0 −3 0 0 −1
 
1 − 2τ −1 + τ 0
−1 + τ −2 − τ 0 <0
0 0 −3 + τ
¿tau?  
2τ − 1 1 − τ 0
⇐⇒  1 − τ 2 + τ 0 >0
0 0 3−τ
Eigenvalue.  
λ + 1 − 2τ −1 + τ 0
 −1 + τ λ−2−τ 0 =
0 0 λ−3+τ

λ1 = 3 − τ
3−τ ≥0
3≥τ
τ ≤3
√
3τ ± 5τ 2 −14τ +13+1
λ2,3 = 2
> 0τ1 = 2,4
√
λ2,3 = 3τ + 1 ± 5τ 2 − 14τ + 13 > 0
A = 3τ + 1
B = 5τ 2 − 14τ + 13
A>B

3
A = 3τ + 1
B = 5τ 2 − 14τ + 13
3τ + 1 − (5τ 2 − 14τ + 13) > 0
3τ + 1 − 5τ 2 + 14τ − 13 > 0
−5τ 2 + 17τ − 12 > 0
5τ 2 − 17τ + 12 < 0
(τ − 2, 4)(τ − 1) ≤ 0
τ1 = 2, 4
τ2 = 1
1 ≤ τ ≤ 2, 4

4
IV. Home task 4:

2 −1 1 0 0 −1
ψ=
, P =
, Q=

−1 2 0 1 −2 0

Has the LMI (!) a solution?

Hint: to ise Corollary to theorem 2.1
    
2 −1 0−2 0 −1
−µ <0
−1 2 −10 −2 0

       
2 −1 4 0 2 − 4µ −1 4µ 1
−µ = <0→
−1 2 0 1 −1 2−µ 1 µ−2

by Sylvester criterian
1
−2 + 4µ > 0 → µ > 2
(−2 + µ)(−2 + µ) − 1 > 0
4 − 10µ + 4µ2 − 1 > 0
4µ − 10µ + 3 > 0
(µ − 2,5114)(µ − 0,3486) > 0

5
 V. Home task 5:
   
−2 1 0
A= , B= , C = (1, 1), D = 2
0 −1 1

x ∈ ℜ2
x∈ℜ

To find the maximun value η ≤ 0 of dissipativity, i.e.

To make itteration

a) η = η0 small enough (η < 1)

b) η = η0 + δ, and solve LMI, if there exist solution then we can follow
c) ηkh = ηk − δ, if no solution then step k give η = ηk as solution

Algoritmo 0.1: Maximun Value of dissipativity

 
−2 1  T  T
Data: A = ,B= 1 0 ,C= 1 1 ,D=2
0 −1
1 x = sdpvar(2);
2 w = sdpvar(1);
3 P = sdpvar(2, 2);
4 ztep = 0,01;
5 err = 0;
6 for n = 0 : ztep : inf ;
′ ′ ′ ′
7 F = [P ∗ A + A ∗ P P ∗ B − C; B ∗ P − C 2 ∗ n ∗ eye(1) − (D + D)] <= 0];
8 F = [F, trace(P ) == 1];
9 t = optimize(F );
10 err = t.problem;
11 if err == 1;
12 break;
13 end;
14 m=n;
15 end;

6
VI. Home task 6: For the set of data from task 5 to check if this system is
expansive or not!

Algoritmo 0.2: The System is expansive or not?

 
−2 1  T  T
Data: A = ,B= 1 0 ,C= 1 1 ,D=2
0 −1
1 x = sdpvar(2);
2 w = sdpvar(1);
3 P = sdpvar(2, 2);
′ ′ ′ ′ ′ ′
4 F = [[P ∗ A + A ∗ P + 2 ∗ C ∗ CP ∗ B + C ∗ D; B ∗ P + D ∗ C − 2 ∗
eye(1) + 2 ∗ D ′ ∗ D] <= 0];
5 F = [F, trace(P ) == 1];
6 t = optimize(F )

The system is not expansive!!

7
VII. Home task 7: For the system from the previous task estimate k H k∞ ≤ γ
(to find a minimal γ )
Hint: make itteration

a) Fix large enough γ and to check if there exist a positve solution P to

the LMI Wγ (P ) ≤ 0
b) If yes, γ1 = γ0 − δ and check again

Algoritmo 0.3: Fin a minimal γ

 
−2 1  T  T
Data: A = ,B= 1 0 ,C= 1 1 ,D=2
0 −1
1 x = sdpvar(2);
2 w = sdpvar(1);
3 P = sdpvar(2, 2);
4 err = 0;
5 for gamma = 3 : −0,01 : 0;
′ 2 ′ ′ 2 ′
6 F = [[P ∗ A + A ∗ P + (2/gamma ) ∗ C ∗ CP ∗ B + (1/gamma ) ∗ C ∗
′ 2 ′ 2 ′
DB ∗ P + (1/gamma ) ∗ D ∗ C − 2 ∗ eye(1) + (2/gamma ) ∗ D ∗ D] <= 0];
7 F = [F, trace(P ) == 1];
8 t = optimize(F );
9 err = t.problem;
10 if err == 1;
11 break
12 end
13 gama=gamma;
14 end
15 gama

8
VIII. Home task 8:

L = 1, xt = Axt + Axt−τ , τ1 = 1seg

   
−2 0 1 2
A= , A1 =
2 1 2 1

To check if this system is stable or not!

Algoritmo 0.4: The system is stable or not?

 
−2 0  T  T
Data: A = ,B= 1 0 ,C= 1 1 ,D=2
0 1
1 P = sdpvar(2, 2);
2 P 1 = sdpvar(2, 2);
′ ′
3 W = [[P ∗ A + A ∗ P + P 1P ∗ A1; A1 ∗ P − P 1] < 0];
4 W = [W, trace(P ) == 1];
5 t = optimize(W );

The system is not stable!

IX. Home task 9:

d
x (t)
dt 1
= a0 x1 (t) + a1 x2 (t − η) ∈ ℜ
x2 (t) = a2 x1 (t) + a3 x2 (t − η) ∈ ℜ
T = 1s, x(θ) = 1 when θ ∈ [−1s, 0]
a2 = 1, a3 = −1
Check if this system is stable!
X. Home task 10:
     
−x1 + 2signx2 1 ξ
A= , B= , ξx = 1x
signalx1 − 2x2 0 ξ2x

To find k providing the minimal attractive ellipsoid convergence

ẋ1 = −x1 + 2signx2 + u + ξ1x

ẋ2 = signx1 − 2x2 + ξ2x
0
ξ1x = ξ1x sin(ωt + ϕ1 )
0
ξ1x = ξ2x cos(ωt + ϕ2 )