Stability Definitions Stability: in the sense of Lyapunov (i.s. L). Asymptotic (Internal) Stability: Zero-input response. Input-Output Stability or Bounded-InputBounded-Output (BIBO) Stability: Zerostate response. Stability, Detectability & Stabilizability M. Sami Fadali Professor of Electrical Engineering University of Nevada 1 Equilibrium State 2 Stability of Equilibrium How sensitive is the system to small perturbations in its equilibrium? Solve = equilibrium state For nonlinear systems, multiple states Stability of equilibrium state: depends on behavior after a perturbation from the equilibrium. Unstable Stable Asymptotically stable 3 4 Stability Stable System Definition: For any there exists a constant such that implies • Can stay arbitrarily close to equilibrium by starting sufficiently close to it. • Unstable: not stable (cannot stay arbitrarily close to the equilibrium. 0 : : = open ball of radius = open ball of radius 5 6 Exponential Stability Asymptotically Stable Definition: There exist positive constants such that implies Definition: Stable equilibrium and it is such that possible to choose implies • Global exponential stability: property holds . for any initial state • Length of state vector decays faster than an exponential function. • For linear systems, decay is always exponential. → • Converges to equilibrium by starting sufficiently close to it. • Globally asymptotically stable: converges to equilibrium from any initial state. 7 8 Exponential Stability Linear Time-invariant Systems 1 • For a nonsingular state matrix if and only if • Only one equilibrium point at the origin. • For a singular state matrix , • Rank deficit=number of linearly independent • Infinitely many equilibrium points on 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 5 10 15 9 10 Asymptotic (Internal) Stability for LTI Systems Stability for LTI Systesm Theorem: LTI system is asymptotically stable if the zero-input response converges to zero for any initial state. • Response is bounded for distinct eigenvalues on the imaginary axis (stable). • Response is unbounded for repeated eigenvalues on the imaginary axis (unstable). 11 • LTI system is asymptotically stable if and only if all eigenvalues are in the open LHP. • LTI: asymptotic stability implies global exponential stability. 12 State Trajectories Example 3 Is the system asymptotically stable? >> A=[3,-2,0;8,-3,-4;0,4,-9]; >> eig(A) ans = -1.0000 -3.0000 -5.0000 Yes. All poles in the open LHP. 2 Stable 1 0 Asymptotically stable -1 -2 -1.5 -1 -0.5 0 0.5 1 13 Example 14 Input-output (BIBO) Stability • Is the system stable? >> A=[-5,5,-5;-20,20,-20;-20,20,-20]; >> eig(A) ans = 0 -5 0 No. Repeated pole on the imaginary axis. Definition For any bounded input, the output is bounded. • Definition can be generalized to time-varying MIMO systems using ||.|| (norm) in place of |.| • Definition can be generalized to distributed parameter systems. 15 16 Contradiction Theorem 1: BIBO Stability Proof Theorem 1 BIBO Stability A SISO LTI system is BIBO stable if and only if its impulse response satisfies • Remarks • Condition can be generalized to time-varying MIMO systems using ||.|| (norm) in place of |.| • Condition can be generalized to distributed parameter systems. • Sufficiency (if) Necessity (only if): Assume BIBO stable with condition violated and let 17 18 Proof: Sufficiency Theorem 2: BIBO Stability , LTI SISO system is BIBO stable if and only if all its transfer function poles are in the open LFP. Proof (Necessity) The integral of diverges for any transfer function pole is in the closed RHP . After pole-zero cancellation, poles =remaining LHP poles (not all ) BIBO stability. 19 20 Relationship Between Internal Stability BIBO Stability Kalman Decomposition U(s) • Any system can be decomposed into four subsystems as shown in the figure: • Unobservable Mode Y(s) • BIBO stability is equivalent to open LHP poles • Internal stability implies BIBO stability (since poles are a subset of the eigenvalues). • Some eigenvalues may cancel in the transfer function and are not poles. • BIBO stability does not , in general, imply internal stability • With no cancellation, {poles}={eigenvalues} BIBO stability is equivalent to internal stability Controllable Observable Y(s) Uncontrollable Observable Us) • Uncontrollable Mode Controllable Unobservable Uncontrollable Unobservable 21 22 Important Relations Definitions x2 Detectable: all unstable modes are observable Unobservable Subspace (stable) (i.e. all unobservable modes are stable). Stabilizable: all unstable x2 modes are controllable Uncontrollable Subspace (i.e. all uncontrollable (stable) modes are stable). Observable Subspace y x1 Controllable Subspace u x1 23 • Internally stable systems are stabilizable and detectable (no unstable modes). • Observable systems are detectable (no unstable unobservable modes). • Controllable systems are stabilizable (no unstable uncontrollable modes). • For minimal realizations, BIBO stability and internal stability are equivalent {poles}={eigenvalues}. 24 Example (continued) Example 1 s 1 s 1 2 s 1s 1 s 1 s 1 1 0 BIBO stable Controllable Form : without cancellation 1 s 1 s 1 2 s 1s 1 s 1 s 1 1 0 BIBO stable Observable Form : without cancellation G( s) G( s) 0 1 0 1 1 A B C 1 1 O 1 0 1 1 1 0 1 1 1 1 A B C 1 0 C 1 0 1 1 1 Controllable (phase var. form) but not observable. BIBO stable but not internally stable. Stabilizable but not detectable. 25 Observable (observer form) but not controllable. BIBO stable but not internally stable. Detectable but not stabilizable. 26