uncertain z w outline well posedness and topological
play

uncertain z + w Outline Well-posedness and topological - PowerPoint PPT Presentation

Quadratic separation for feedback connection of an uncertain matrix and an implicit linear transformation Dimitri PEAUCELLE & Didier HENRION & Denis ARZELIER all with LAAS-CNRS - Toulouse, FRANCE also with Czech Technical


  1. Quadratic separation for feedback connection of an uncertain matrix and an implicit linear transformation Dimitri PEAUCELLE & Didier HENRION ‡ & Denis ARZELIER all with LAAS-CNRS - Toulouse, FRANCE ‡ also with Czech Technical University in Prague

  2. Problem statement Well-posedness of interconnected system + A , E possibly non square z w ∇ ∈ ∇ ∇ uncertain z + w Outline ① Well-posedness and topological separation ② Robust stability of descriptor systems ③ Main result ④ Some results for robust analysis ⑤ Numerical example 1 GT MOSAR 9-10 juin 2005

  3. Topological separation General framework w G (z, w)=0 w Well-Posedness: z w Bounded ( ¯ w, ¯ z ) ⇒ unique bounded ( w, z ) F (w, z)=0 z z ∃ θ topological separator: F (¯ z ) = { ( w, z ) : F ¯ z ( w, z ) = 0 } ⊂ { ( w, z ) : θ ( w, z ) > − φ 1 ( || ¯ z || ) } G I ( ¯ w ) = { ( w, z ) : G ¯ w ( z, w ) = 0 } ⊂ { ( w, z ) : θ ( w, z ) ≤ φ 2 ( || ¯ w || ) } Related results : ➞ Stability ( θ Lyapunov certificate), Passivity ( θ storage function), IQC ... ➞ Robust analysis of Linear uncertain systems (Iwasaki) 2 GT MOSAR 9-10 juin 2005

  4. Well-posedness in the considered case + Well-posedness: z w Null ( E − A∇ ) is empty ∀∇ ∈ ∇ ∇ z + w ✪ Includes classical µ theory framework: I − A∇ non-singular for all (structured) norm-bounded ∇ ✪ Results of the paper extend to block diagonal uncertainties: ∇ = diag ( δ R 1 I r 1 , . . . , δ R N R I r N R , δ C 1 I c 1 , . . . , δ C N C I c N C , ∆ C 1 , . . . , ∆ C N f ) 3 GT MOSAR 9-10 juin 2005

  5. Well-posedness and descriptor systems ⇔ Es − A full rank for all s ∈ C + Stability of E ˙ x = Ax   w = ∇ z  for all ∇ = s − 1 I n ∈ C + ⇔ W.P . of Ez = Aw   + x x + 4 GT MOSAR 9-10 juin 2005

  6. Well-posedness and pole location Let a region defined as a half plane or a disk 2 s ∗ + d 3 ss ∗ ≤ 0 } D = { s ∈ C : d 1 + d 2 s + d ∗ D -Stability of E ˙ x = Ax ⇔ Es − A full rank for all s ∈ D   w = ∇ z  ⇔ W.P for all ∇ = ∇ ∇ . of  Ez = Aw  ∇ = { s − 1 I n : d 1 s − 1 s −∗ + d 2 s −∗ + d ∗ 2 s − 1 + d 3 ≥ 0 } . where ∇ 5 GT MOSAR 9-10 juin 2005

  7. Well-posedness and polynomial descriptor systems Stability of A d x ( d ) + A d − 1 x ( d − 1) + · · · + A 1 ˙ x + A 0 x = 0   w = ∇ z  for all ∇ = s − 1 I d n ∈ C + ⇔ W.P . of E z = A w       · · · · · · A d O O A d − 1 A 1 A 0         − I O O I O O         where E = A = − . .     ... ... . .     . .         O O − I O I O 6 GT MOSAR 9-10 juin 2005

  8. Robust stability of descriptor systems Robust stability ∀ ∆ ∈ ∆ ∆ of E (∆) ˙ x = A (∆) x with E (∆) = E A + ( B ∆ − E B )( E D − D ∆) − 1 E C A (∆) = A + ( B ∆ − E B )( E D − D ∆) − 1 C    s − 1 ∈ C +  s − 1 I n  w = ∇ z O    ⇔ W.P for all ∇ = : . of   E z = A w O ∆ ∆ ∈ ∆ ∆       E A E B  A B     where E = A =   E C E D C D 7 GT MOSAR 9-10 juin 2005

  9. Main result: Quadratic separation   w = ∇ z  for all ∇ = ∇ ∇ W.P . of  E z = A w     � � I    E ◦∗ ∇ ∗    ≤ O , ∀∇ ∈ ∇ ∇ Θ I      ∇E ◦ ⇔ ∃ Θ :   � � ⊥∗ � � ⊥    EE ◦ EE ◦ > O .  Θ −A −A  where the columns of E ◦ form an orthogonal basis of E ∗ � � ⊥ � � EE ◦ EE ◦ −A −A and the columns of span the null-space of . 8 GT MOSAR 9-10 juin 2005

  10. Application to descriptor systems   w = ∇ z  for all ∇ = s − 1 I n ∈ C + . Stability of E ˙ x = Ax ⇔ W.P . of Ez = Aw   Choice of separator: Θ � �� �     − E ◦∗ P � � O I  = − 2 Re ( s − 1 ) E ◦∗ PE ◦ E ◦∗ s −∗     I    E ◦ s − 1 − PE ◦ O E ◦∗ PE ◦ > O   LMI result ❶ : E ◦∗ P � � ⊥∗ O � � ⊥ EE ◦   EE ◦ < O − A − A   PE ◦ O E ∗ X ∗ = XE ≥ O , A ∗ X ∗ + XA < O Equivalent to ❷ : 9 GT MOSAR 9-10 juin 2005

  11. Example of descriptor system      1  1    ˙ x =    x Scalar ’switch’ system: 0 α ● x ( t ) = 0 the system is stable If α � = 0 � � ⊥ ❶ EE ◦ = [ ] → LMI p > 0 − A � � ❷ X = → LMI x 1 ≥ 0 , 2 x 1 + 2 x 2 α < 0 x 1 x 2 ● ˙ If α = 0 x = x the system is unstable    1 � � ⊥ ❶ =   → LMI p > 0 , 2 p < 0 EE ◦ − A  1 � � ❷ X = → LMI x 1 ≥ 0 , 2 x 1 < 0 x 1 x 2 10 GT MOSAR 9-10 juin 2005

  12. Application to uncertain descriptor systems : | δ | ≤ ¯ Robust Stability of E (∆) ˙ x = A (∆) x , ∆ = δ I δ    s − 1 ∈ C +  s − 1 I n  w = ∇ z O    ⇔ for all ∇ = : W.P . of .  | δ | ≤ ¯ E z = A w  O δ I m δ  Choice of separator:   E ◦∗ 2 Θ 1 E ◦ −E ◦∗ E ◦∗ 1 P 2 Θ 2   2  s − 1 E ◦   : ∇E ◦ =   1 Θ =   − P E ◦ O O    1   ∆ E ◦   2 Θ 2 ∗ E ◦ O Θ 3 2  E ◦∗ 1 P E ◦ E ◦∗ 2 Θ 3 E ◦  1 ≥ O , 2 ≥ O       δ Θ 2 ∗ + ¯ 2 ( Θ 1 + ¯ δ Θ 2 + ¯ E ◦∗ δ 2 Θ 3 ) E ◦ 2 ≤ O constrained by the LMIs     δ Θ 2 ∗ + ¯ 2 ( Θ 1 − ¯ δ Θ 2 − ¯  δ 2 Θ 3 ) E ◦ E ◦∗  2 ≤ O  11 GT MOSAR 9-10 juin 2005

  13. Reducing conservatism ◆ Original system            s − 1 I n  ˙  E A E B x  A B  x O      =      , ∇ =      E C E D z ∆ C D w ∆ O δ I m ❖ Augmented system     O I O O I O O O         ¨ ˙ x x     O E A E B O O A B O                 ˙ x x                 = O E C E D O O C D O                 z ∆ w ∆         E A O O E B A O O B             ˙ ˙ z ∆ w ∆     E C O O E D C O O D    s − 1 I 2 n O   , ∇ =  O δ I 2 m 12 GT MOSAR 9-10 juin 2005

  14. Reducing conservatism ◆ Quadratic separation on Original system ➞ Lyapunov stability with V ( η, ∆) = η ∗ Pη ❖ Quadratic separation on Augmented system ➞ Lyapunov stability with V ( η, ∆) = η ∗ P (∆) η where in the case E = I (not descriptor)     ∗  A (∆)  A (∆)     P (∆) = P   I I 13 GT MOSAR 9-10 juin 2005

  15. Robust stability analysis example     − 16 0 0 8 0 0 0 0 3 0 0 0         − 8 − 16 0 0 0 0 4 8 0 9 2 0             − 19 − 17 0 0 10 9 0 0 0 0 0 0         E = , A =        − 20  10 0 0 0 0 0 0 0 1 4 0             − 10 0 3 0 0 0 5 0 1 0 2 0         − 6 − 6 3 1 0 0 3 0 3 0 7 0 ◆ Quadratic separation on Original system ➞ LMIs feasible up to ¯ δ = 0 . 22 - infeasible for ¯ δ = 0 . 23 ❖ Quadratic separation on Augmented system ➞ LMIs feasible up to ¯ δ = 0 . 45 - infeasible for ¯ δ = 0 . 46 14 GT MOSAR 9-10 juin 2005

  16. Conclusions ➞ Quadratic separation for feedback connection of an uncertain matrix and an implicit linear transformation is valuable for extensions of known results to de- scriptor systems. ➞ Applying existing methods to artificially augmented systems gives new less conservative results (more about this in Seville) 15 GT MOSAR 9-10 juin 2005

Recommend


More recommend