Integral Quadratic Separation Framework Dimitri PEAUCELLE LAAS-CNRS - - PowerPoint PPT Presentation

Integral Quadratic Separation Framework

Dimitri PEAUCELLE LAAS-CNRS - Universit´ e de Toulouse - FRANCE joint work with Denis Arzelier and Fr´ ed´ eric Gouaisbaut Seminar March 2011

Outline ➊ Topological separation & related theory

• Well-posedness definition and main result
• Relations with Lyapunov theory
• The case of linear uncertain systems : quadratic separation
• The Lur’e problem
• Relations with IQC framework

➋ Integral Quadratic Separation (IQS) for the descriptor case ➌ Performances in the IQS framework ➍ System augmentation : a way towards conservatism reduction ➎ The Romuald toolbox

1 March 2011, Unicamp

➊ Topological separation & related theory

■ Well-posedness

G (z, w)=0

z w

z z w w F (w, z)=0

Well-Posedness: Bounded ( ¯

w, ¯ z) ⇒ unique bounded (w, z)

• In case

F

• z = Aw + ¯

z and

G

• w = ∆z + ¯

w are linear applications

Well-posedness : (1 − A∆) non-singular

▲ What if ∆ = ∆ ∈ ∆ ∆ is uncertain ? ▲ If A = T(jω) is an LTI system ? ▲ If G is non-linear ?

...

2 March 2011, Unicamp

➊ Topological separation & related theory

■ Well-posedness & topological separation

G (z, w)=0

z w

z z w w F (w, z)=0

Well-Posedness: Bounded ( ¯

w, ¯ z) ⇒ ∃!(w, z) , ∃γ :

• w

z

• ≤ γ
• ¯

w ¯ z

• [Safonov 80] ∃θ topological separator:

GI( ¯ w) = {(w, z) : G ¯

w(z, w) = 0} ⊂ {(w, z) : θ(w, z) ≤ φ2(|| ¯

w||)} F(¯ z) = {(w, z) : F¯

z(w, z) = 0} ⊂ {(w, z) : θ(w, z) > −φ1(||¯

z||)}

• φ1 and φ2 are positive functions. When ¯

w = 0, ¯ z = 0 separation reads as GI(0) = {(w, z) : G0(z, w) = 0} ⊂ {(w, z) : θ(w, z) ≤ 0} F(0) = {(w, z) : F0(w, z) = 0} ⊂ {(w, z) : θ(w, z) > 0}

3 March 2011, Unicamp

➊ Topological separation & related theory

■ For dynamic systems ˙ x = f(x), topological separation ≡ Lyapunov theory

F

• z(t) = f(w(t)) + ¯

z(t) ,

G

• w(t)
• x(t)

= t z(τ)

• ˙

x(t)

dτ + ¯ w(t) ▲ ¯ w : contains information on initial conditions (x(0) = 0 by convention)

• Well-posedness ⇒ for zero initial conditions and zero perturbations :

w = z = 0 (equilibrium point).

• Well-posedness (global stability)

⇒ whatever bounded perturbations the state remains close to equilibrium

4 March 2011, Unicamp

➊ Topological separation & related theory

■ For dynamic systems ˙ x = f(x), topological separation ≡ Lyapunov theory

F

• z(t) = f(w(t)) + ¯

z(t) ,

G

• w(t)
• x(t)

= t z(τ)

• ˙

x(t)

dτ + ¯ w(t)

• Assume a Lyapunov function V (0) = 0 , V (x) > 0 , ˙

V (x) < 0 ▲ Topological separation w.r.t. GI(0) is obtained with θ(w = x, z = ˙ x) = ∞ −∂V ∂x (x(τ)) ˙ x(τ)dτ = lim

t→∞ −V (x(t)) < 0

▲ Topological separation w.r.t. F(0) does hold as well θ(w, z = f(w)) = ∞ − ˙ V (w(τ))dτ > 0

5 March 2011, Unicamp

➊ Topological separation & related theory

■ For linear systems : quadratic Lyapunov function, i.e. quadratic separator

F

• z(t) = Aw(t) + ¯

z(t) ,

G

• w(t)
• x(t)

= t z(τ)

• ˙

x(t)

dτ + ¯ w(t)

• A possible separator based on quadratic Lyapunov function V (x) = xTPx

θ(w, z) = ∞

• zT (τ)

wT (τ) −P − P z(τ) w(τ)

• dτ

▲ Quadratic separation w.r.t. GI(0): lim

t→∞ −xT(t)Px(t) ≤ 0 , i.e. P > 0

▲ Quadratic separation w.r.t. F(0) guaranteed if ∀t > 0 , − 2wT(t)PAw(t) > 0 , i.e. ATP + PA < 0

6 March 2011, Unicamp

➊ Topological separation & related theory

■ Topological separation : alternative to Lyapunov theory ▲ Needs to manipulate systems in a new form

• Suited for systems described as feedback connected blocs

Any linear system with rational dependence w.r.t. parameters writes as such

˙ x = (A + B∆∆(1 − D∆∆)−1C∆)x

LFT

← →            ˙ x = Ax + B∆w∆ z∆ = C∆x + D∆w∆ w∆ = ∆z∆ ▲ Finding a topological separator is a priori

as complicated as finding a Lyapunov function

• Allows to deal with several features simultaneously in a unified way

7 March 2011, Unicamp

➊ Topological separation & related theory

■ Quadratic separation [Iwasaki & Hara 1998]

• If F(w) = Aw is a linear transformation

and G = ∆ is an uncertain operator defined as ∆ ∈ ∆

∆ convex set

it is necessary and sufficient to look for a quadratic separator

θ(z, w) = ∞

• zT

wT

• Θ
• z

w

• dτ
• If F(w) = A(ω)w is a linear parameter dependent transformation

and G = ∆ is an uncertain operator defined as ∆ ∈ ∆

∆ convex set

necessary and sufficient to look for a parameter-dependent quadratic separator

θ(z, w) = ∞

• zT

wT

• Θ(ω)
• z

w

• dτ

8 March 2011, Unicamp

➊ Topological separation & related theory

■ A well-known example : the Lur’e problem

G (z, w)=0

z w

z z w w F (w, z)=0

▲ F = T(jω) is a transfer function ▲ G(z)/z ∈ [ − k1, − k2 ] is a sector-bounded gain

(i.e. the inverse graph of G is in [ − 1/k1 , − 1/k2 ])

• Circle criterion : exists a quadratic separator (circle) for all ω

9 March 2011, Unicamp

➊ Topological separation & related theory

■ Another example : parameter-dependent Lyapunov function

G (z, w)=0

z w

z z w w F (w, z)=0

▲ F = A(δ) parameter-dependent LTI state-space model ▲ G = I is an integrator

• Necessary and sufficient to have

Θ(δ) =   −P(δ) − P(δ)  

10 March 2011, Unicamp

➊ Topological separation & related theory

■ Direct relation with the IQC framework ▲ F = T(jω) is a transfer matrix ▲ G = ∆ is an operator known to satisfy an Integral Quadratic Constraint (IQC) +∞

−∞

• 1

∆∗(jω)

• Π(ω)

  1 ∆(jω)   dω ≤ 0

• Stability of the closed-loop is guaranteed if for all ω
• T ∗(jω)

1

• Π(ω)

  T(jω) 1   > 0 ▲ Knowing ∆ ∆ how to choose Π = Θ? (i.e. the quadratic separator)

Plenty of results in µ-analysis and IQC theory D-scalings, DG-scalings etc. but still, conservative

11 March 2011, Unicamp

Outline ➊ Topological separation & related theory ➋ Integral Quadratic Separation (IQS) for the descriptor case ➌ Performances in the IQS framework ➍ System augmentation : a way towards conservatism reduction ➎ The Romuald toolbox

12 March 2011, Unicamp

➋ IQS for the descriptor case

G (z, w)=0

z w

z z w w F (w, z)=0

z w w z

■ Linear implicit application in feedback loop with an uncertain operator Ez(t) = Aw(t)

• F

, w(t) = [∇z](t)

• G

∇ ∈ ∇ ∇

• ∇ is bloc-diagonal contains scalar, full-bloc, LTI and LTV uncertainties

and other operators such as integrators, delays...

• Result also extend to time-varying linear applications E(t), A(t)

and to polytopic linear applications

• E(ξ)

A(ξ)

• ∈ CO
• E[i]

A[i]

• .

13 March 2011, Unicamp

➋ IQS for the descriptor case

■ Integral Quadratic Separation [Automatica’08, CDC’08]

• For the case of linear application with uncertain operator

Ez(t) = Aw(t) , w(t) = [∇z](t) ∇ ∈ ∇ ∇

where E = E1E2 with E1 full column rank,

• Integral Quadratic Separator (IQS) : ∃Θ, matrix, solution of LMI
• E1

−A ⊥∗ Θ

• E1

−A ⊥ > 0

and Integral Quadratic Constraint (IQC) ∀∇ ∈ ∇

∇ ∞   E2z(t) [∇z](t)  

∗

Θ   E2z(t) [∇z](t)   dt ≤ 0

14 March 2011, Unicamp

➋ IQS for the descriptor case

• For given ∇

∇, there exist (conservative) LMI conditions for Θ solution to IQC ∞   E2z(t) [∇z](t)  

∗

Θ   E2z(t) [∇z](t)   dt ≤ 0 ▲ Θ is build out of IQS for elementary blocs of ∇ ▲ Improved DG-scalings, full-bloc S-procedure, vertex separators... ▲ Building Θ and related LMIs is tedious but can be automatized www.laas.fr/OLOCEP/romuloc/ ▲ It is conservative except in few special cases [Meinsma et al., 1997].

15 March 2011, Unicamp

➋ IQS for the descriptor case

■ Robust analysis in IQS framework:

• 1- Write the robust analysis problem as a well-posedness problem

Ez = Aw , w = ∇z =     ∇1

...

∇¯



    z

• 2- Build Integral Quadratic Separators for each elementary bloc ∇j
• 3- Apply the IQS results to get (conservative) LMIs

16 March 2011, Unicamp

Outline ➊ Topological separation & related theory ➋ Integral Quadratic Separation (IQS) for the descriptor case ➌ Performances in the IQS framework ➍ System augmentation : a way towards conservatism reduction ➎ The Romuald toolbox

17 March 2011, Unicamp

➌ Performance analysis in quadratic separation framework

■ Induced L2 norm (H∞ in the LTI case) E ˙ x = Ax + Bv , g = Cx + Dv ▲ Prove that system is asymptotically stable ▲ and g < γv for zero initial conditions x(0) = 0

(strict upper bound on the L2 gain attenuation)

• Equivalent to well-posedness with respect to

Integrator with zero initial conditions x(t) = [I1 ˙

x](t) = t

0 ˙

x(τ)dτ

and signals such that v ≤ 1

γg 18 March 2011, Unicamp

➌ Performance analysis in quadratic separation framework

■ Induced L2 norm E ˙ x = Ax + Bv , g = Cx + Dv ▲ Define ∇n2n the fictitious non-causal uncertain operator such that v = ∇n2ng

iff v ≤ 1

γ g

• Induced L2 norm problem is equivalent to well-posedness of

  E 1  

• E

  ˙ x g  

z

=   A B C D  

• A

  x v  

w

, ∇ =   I1 ∇n2n  

19 March 2011, Unicamp

➌ Performance analysis in quadratic separation framework

■ Induced L2 norm   E 1  

• E

  ˙ x g  

z

=   A B C D  

• A

  x v  

w

, ∇ =   I1 ∇n2n  

• Elementary IQS for bloc I1 is

ΘI1 =   −P −P   : P > 0

Indeed (recall x(t) = [I1 ˙

x](t) = t

0 ˙

x(τ)dτ and x(0) = 0)   ˙ x I1 ˙ x  

• ΘI1

  ˙ x I1 ˙ x  

• T

= −x∗(T)Px(T) ≤ 0

20 March 2011, Unicamp

➌ Performance analysis in quadratic separation framework

■ Induced L2 norm   E 1  

• E

  ˙ x g  

z

=   A B C D  

• A

  x v  

w

, ∇ =   I1 ∇n2n  

• Elementary IQS for bloc ∇n2n is (small gain theorem)

Θ∇n2n =   −τ1 τγ21   : τ > 0

Indeed (recall v = ∇n2ng iff v ≤ 1

γg)

  g ∇n2ng  

• Θ∇n2n

  g ∇n2ng  

• = τ(−g2 + γ2v2) ≤ 0

21 March 2011, Unicamp

➌ Performance analysis in quadratic separation framework

• Apply IQS and get (for non-descriptor case E = 1)

P > 0 , τ > 0   A∗P + PA + τC∗C PB + τC∗D B∗P + τD∗C −τγ21 + τD∗D   < 0

which is the classical H∞ result.

• No difficulty to generate LMIs for descriptor case
• No difficulty to handle systems with uncertainties, time-delays...

22 March 2011, Unicamp

➌ Performance analysis in quadratic separation framework

■ Generic robust performance analysis problem:

• Well-posedness of

▲

• 1n integrator

▲ ∆ matrix of uncertainties ▲ ∇perf operator related to performances

(induced L2, H∞, H2, impulse-to-norm, norm-to-peak, impulse-to-peak)

23 March 2011, Unicamp

Outline ➊ Topological separation & related theory ➋ Integral Quadratic Separation (IQS) for the descriptor case ➌ Performances in the IQS framework ➍ System augmentation : a way towards conservatism reduction ➎ The Romuald toolbox

24 March 2011, Unicamp

➍ System augmentation and conservatism reduction

■ Towards less-conservative conditions: System augmentation ▲ Example of stability of uncertain system with parametric uncertainty (˙ δ = 0) ˙ x = (A + δB∆(1 − δD∆)−1C∆)x ▲ Corresponds to well-posedness of   ˙ x z∆   =   A B∆ C∆ D∆     x w∆   , ∇ =   I11n δ1m   ▲ [Meinsma] rule indicates results may be conservative

25 March 2011, Unicamp

➍ System augmentation and conservatism reduction

▲ Well-posedness of   ˙ x z∆   =   A B∆ C∆ D∆     x w∆   , ∇ =   I1n δ1m  

• adding the fact that ˙

w∆ = δ ˙ z∆, is also equivalent to well-posedness of         

1 1 1 −C∆ 1

               

˙ w∆ ˙ x z∆ ˙ z∆

       =         

1 A B∆ C∆ D∆ D∆ 1 −1

               

w∆ x w∆ ˙ w∆

       ∇ =   I1m+n δ12m  

26 March 2011, Unicamp

➍ System augmentation and conservatism reduction

        

1 1 1 −C∆ 1

               

˙ w∆ ˙ x z∆ ˙ z∆

       =         

1 A B∆ C∆ D∆ D∆ 1 −1

               

w∆ x w∆ ˙ w∆

       ∇ =   I1m+n δ12m   ▲ It is descriptor model.

• More decisions variables in the separator (increased dimensions of ∇)
• Bigger LMI conditions (m + n rows)

27 March 2011, Unicamp

➍ System augmentation and conservatism reduction

• Lyapunov function is with respect to the augmented state

(vector involved in the integrator operator)

• w∗

∆

x∗

• P

  w∆ x   ▲ Recalling that w∆ = δ(1 − δD∆)−1C∆x

the result corresponds to looking for a parameter dependent Lyapunov function

x∗   δ(1 − δD∆)−1C∆ 1  

∗

P   δ(1 − δD∆)−1C∆ 1   x

• Proves to be less conservative than for LMIs obtained on original system.

28 March 2011, Unicamp

➍ System augmentation and conservatism reduction

■ Towards less-conservative conditions: System augmentation

• Adding more equations for higher derivatives of the state:

less conservative LMI conditions

• Same technique works for time varying uncertainties

(if known bounds on derivatives)

• Has been applied successfully to time-delay systems [Gouaisbaut]:

gives sequences of LMI conditions with decreasing conservatism

▲ Related to SOS representations of positive polynomials [Sato 2009]:

conservatism decreases as the order of the representation is augmented

• No need to manipulate by hand LMIs (Schur complements etc.), polynomials...

▲ Does conservatism vanishes? Exactly? Asymptotically? ▲ Is it possible to cope with non-linearities in the same way?

29 March 2011, Unicamp

Outline ➊ Topological separation & related theory ➋ Integral Quadratic Separation (IQS) for the descriptor case ➌ Performances in the IQS framework ➍ System augmentation : a way towards conservatism reduction ➎ The Romuald toolbox

30 March 2011, Unicamp

➎ The Romuald toolbox

■ Freely distributed software to test the theoretical results

• Existing software : RoMulOC

www.laas.fr/OLOCEP/romuloc ▲ Contains some of the analysis results plus some state-feedback features

• Currently developed software : Romuald

▲ Dedicated to analysis of descriptor systems ▲ Fully coded using the quadratic separation theory ▲ Allows systematic system augmentation ▲ First preliminary tests currently done for satellite and plane applications

>> quiz = ctrpb( OrderOfAugmentation ) + h2 (usys); >> result = solvesdp( quiz )

31 March 2011, Unicamp

➎ The Romuald toolbox

32 March 2011, Unicamp

➎ The Romuald toolbox

33 March 2011, Unicamp

➎ The Romuald toolbox

34 March 2011, Unicamp

➎ The Romuald toolbox

35 March 2011, Unicamp

Conclusions

36 March 2011, Unicamp