stability of nonlinear systems

Upload: amgadasalama239

Post on 09-Apr-2018

243 views

Category:

Documents


1 download

TRANSCRIPT

  • 8/7/2019 Stability of Nonlinear Systems

    1/34

    Stability of Nonlinear Systems

    By

    Lyapunov Stability

    1

  • 8/7/2019 Stability of Nonlinear Systems

    2/34

    Introduction to Stability

    1. The concept of stability2. Critical points

    3. Linear stability analysis

    2

  • 8/7/2019 Stability of Nonlinear Systems

    3/34

    The Concept of Stability

    Imprecise definition

    Consider a nonlinear system with the origin as a steady-state point:

    Does the system return to the origin if perturbed away from theorigin? If so, the system is stable. Otherwise, the system is unstable.

    0yfyfy

    !! )()(dt

    d

    0yyyfy

    !!!gp

    )(lim)0()( tdt

    d

    t

    H

    I

    y(0)y(t) 0

    y1

    y2

    Precise definition

    Stability: produce a bound I on y(0)

    such that y(t) remains within a givenbound H

    Asymptotic stability: stable & y(t)

    converges to the origin

    Commonly known as Lyapunov

    stability3

  • 8/7/2019 Stability of Nonlinear Systems

    4/34

    Critical Points of a Linear System

    Two-dimensional system

    Divide equations

    Critical point

    Point where dy2/dy1 becomes undetermined Only the origin for a homogeneous linear system

    Five types of critical points depending on the geometric shape of

    trajectories near the origin and eigenvalues ofA matrix

    222121

    2

    212111

    1

    yayadt

    dy

    yayadt

    dy

    dt

    d

    !

    !! Ax

    y

    212111

    222121

    1

    2

    1

    2

    yaya

    yaya

    dtdy

    dtdy

    dy

    dy

    !!

    4

  • 8/7/2019 Stability of Nonlinear Systems

    5/34

    Types of Critical Points

    Proper node

    Two identical realeigenvalues

    Improper node Two different real

    eigenvalues

    Saddle point Two real eigenvalues

    with different signs

    Center

    Two imaginaryeigenvalues

    Spiral point Two complex

    eigenvalues

    Degeneratenode

    No eigenvectorbasis exists

    5

  • 8/7/2019 Stability of Nonlinear Systems

    6/34

    Linear Stability Analysis

    General solution formfor distinct eigenvalues

    Procedure

    Compute theeigenvalues ofA

    The system isasymptotically stable ifand only if Re(Pi) < 0 fori= 1, 2, , n

    The origin is unstable ifRe(Pi) > 0 for any i

    Stability allows zeroeigenvalues

    Imaginary

    RealStable

    Region

    Unstable

    Region

    Left-Half Plane Right-Half Plane

    0

    tn

    n

    tt nececectPPP )()2(

    2

    )1(

    1

    21)( xxxy ! .

    6

  • 8/7/2019 Stability of Nonlinear Systems

    7/34

    Nonlinear Systems

    Steady-state points

    Nonlinear models can have multiple steady states

    Stability must be determined for each steady state

    Consider origin as a generic steady-state point

    Nonlinear model linearization about origin

    00gygy

    ygyyfyy

    yyy

    0yfyf

    y

    !d!d

    d|d!!d

    !

    !!

    )()(

    )()('

    )()(

    dt

    ddt

    d

    dt

    ddt

    d

    yAy

    yfy

    d!d

    !dt

    d

    dt

    d)(

    7

  • 8/7/2019 Stability of Nonlinear Systems

    8/34

    Lyapunov Stability

    8

  • 8/7/2019 Stability of Nonlinear Systems

    9/34

    x0

    p1(0)

    t - time flow

    p2(t)

    p1(t)

    p2(0)

    x(t)

    Definition of Lyapunov Exponents

    Given a continuous dynamical system in an n-dimensional

    phase space, we monitor the long-term evolution of an

    infinitesimal n-sphere of initial conditions.

    The sphere will become an n-ellipsoid due to the locally

    deforming nature of the flow.

    The i-th one-dimensional Lyapunov exponent is then defined

    as following:

    9

  • 8/7/2019 Stability of Nonlinear Systems

    10/34

    On more formal level

    The Multiplicative Ergodic Theorem of Oseledec states thatThe Multiplicative Ergodic Theorem of Oseledec states that

    this limit exists for almost all points xthis limit exists for almost all points x00 and almost alland almost all

    directions of infinitesimal displacementdirections of infinitesimal displacement in the same basin ofin the same basin of

    attraction.attraction. 10

  • 8/7/2019 Stability of Nonlinear Systems

    11/34

    Order: 1> 2 >> n

    The linear extent of the ellipsoid grows as 2

    1t

    The area defined by the first 2 principle axes grows

    as 2(1+2)t

    The volume defined by the first 3 principle axes

    grows as 2(1+2+3)t and so on The sum of the firstjexponents is defined by the

    long-term exponential growth rate of a j-volumeelement.

    11

  • 8/7/2019 Stability of Nonlinear Systems

    12/34

    Signs of the Lyapunov exponents

    Any continuous time-dependent DS without afixed point will have u1 zero exponents.

    The sum of the Lyapunov exponents must be

    negative in dissipative DS at least onenegative Lyapunov exponent.

    A positive Lyapunov exponent reflects adirection ofstretching andfolding and

    therefore determines chaos in the system.

    12

  • 8/7/2019 Stability of Nonlinear Systems

    13/34

    The signs of the Lyapunov exponents provide a

    qualitative picture of a systems dynamics

    1D maps: ! 1=:

    =0 a marginally stable orbit;

    0 chaos.

    3D continuous dissipative DS: (1,2,3)

    (+,0,-) a strange attractor;

    (0,0,-) a two-torus; (0,-,-) a limit cycle;

    (-,-,-) a fixed point.

    13

  • 8/7/2019 Stability of Nonlinear Systems

    14/34

    The sign of the Lyapunov Exponent

    P0- the system is chaotic andunstable. Nearby points willdiverge irrespective of how close

    they are. 14

  • 8/7/2019 Stability of Nonlinear Systems

    15/34

    Computation of Lyapunov Exponents

    Obtaining the Lyapunov exponents from a system

    with known differential equations is no real problem

    and was dealt with by Wolf. In most real world situations we do not know the

    differential equations and so we must calculate the

    exponents from a time series of experimental data.

    Extracting exponents from a time series is a complexproblem and requires care in its application and the

    interpretation of its results.

    15

  • 8/7/2019 Stability of Nonlinear Systems

    16/34

    Calculation of Lyapunov spectra from

    ODE (Wolf et al.) A fiducial trajectory (the center of the sphere) is

    defined by the action of nonlinear equations ofmotions on some initial condition.

    Trajectories of points on the surface of the sphereare defined by the action of linearized equations onpoints infinitesimally separated from the fiducialtrajectory.

    Thus the principle axis are defined by the evolutionvia linearized equations of an initially orthonormalvector frame {e1,e2,,en} attached to the fiducialtrajectory

    16

  • 8/7/2019 Stability of Nonlinear Systems

    17/34

    Problems in implementing:

    Principal axis diverge in magnitude.

    In a chaotic system each vector tends to fall

    along the local direction of most rapid growth.

    (Due to the finite precision of computer calculations, thecollapse toward a common direction causes the tangent space

    orientation of all axis vectors to become indistinguishable.)

    Solution:Gram-Schmidt reorthonormalization (GSR)

    procedure!

    17

  • 8/7/2019 Stability of Nonlinear Systems

    18/34

    GSR never affects the direction of the first vector, sov

    1

    tends to seek out the direction in tangent space

    which is most rapidly growing, |v1|~ 21t;

    v2 has its component along v1 removed and then isnormalized, so v2 is not free to seek for direction,however

    {v1,v2} span the same 2D subspace as {v1,v2}, thusthis space continually seeks out the 2D subspace that

    is most rapidly growing|S(v1,v2)|~2(1+2)t

    |S(v1,v2,vk)|~2(1+2++k)t

    k-volume So monitoring k-volume growth we can find first k

    Lyapunov exponents.

    18

  • 8/7/2019 Stability of Nonlinear Systems

    19/34

    Lyapunov spectrum for experimental data

    (Wolf et al.)

    Experimental data usually consist of discrete

    measurements of a single observable.

    Need to reconstruct phase space with delay

    coordinates and to obtain from such a time series an

    attractor whose Lyapunov spectrum is identical tothat of the original one.

    19

  • 8/7/2019 Stability of Nonlinear Systems

    20/34

    Procedure for P1

    20

  • 8/7/2019 Stability of Nonlinear Systems

    21/34

    Procedure for P1+P2

    21

  • 8/7/2019 Stability of Nonlinear Systems

    22/34

    Lyapunov Simple Example

    A 2D mapf:R2 R2.

    (from Mathworld)

    Define a Lyapunov function.

    The derivative is negative sothe origin is stable.

    212

    21

    2)(

    )(

    xxxf

    xxf

    !

    !

    )()()(* xfxVxV !

    2/)()( 22

    2

    1xxxV !

    2

    2

    *

    21221

    *

    2)(

    )2()(

    xxV

    xxxxxxV

    !

    !

    22

  • 8/7/2019 Stability of Nonlinear Systems

    23/34

    Matlab Tools for Stability Analysis

    Matlab provides several functions forlinear stability analysis of nonlinearsystems

    fsolve finds steady-state point fornonlinear ODE system

    linmod linearizes nonlinear ODE systemabout given steady state to generate

    linear ODE system

    Eigenvalue computes eigenvalues oflinear ODE system

    23

  • 8/7/2019 Stability of Nonlinear Systems

    24/34

    Controllability

    Observability

    )()(

    stateseduncontroll%

    ),(

    corankAlengthunco

    BActrbco

    !

    !

    )()(statesunobserved

    ),(

    obranklengthunob

    Cobsvob

    !

    !

    24

  • 8/7/2019 Stability of Nonlinear Systems

    25/34

    Example

    ? A

    -

    !

    -

    -

    -

    !

    -

    2

    1

    1

    2

    1

    2

    1

    04761.09558.1

    0

    1

    00415.50

    0415.501847.7

    x

    xy

    ux

    x

    x

    x

    Model equations

    Controllability

    2)(;50.04150

    7.1847-1.0),(

    ];0;1[

    0];50.0415;0415.501847.7[

    !

    -

    !!

    !

    !

    corankctrbco

    2)(;97.8712-16.4343-

    0.0476-1.9558),(

    0.04761];-1.9558[

    0];50.0415;0415.501847.7[

    !

    -

    !!

    !

    !

    obrankCobsvob

    CObservability

    25

  • 8/7/2019 Stability of Nonlinear Systems

    26/34

    Example

    ? A

    -

    !

    -

    -

    -

    !

    -

    2

    1

    1

    2

    1

    2

    1

    04761.09558.1

    0

    1

    00415.50

    0415.501847.7

    x

    xy

    ux

    x

    x

    x

    Model equations

    Analysis

    i

    i

    Aeig

    A

    49.91243.5924-49

    .912

    43.59

    24-

    )(

    0];50.0415;0415.501847.7[

    !

    asymptotically stable because all the eigenvalues

    of the state matrix A have negative real parts

    26

  • 8/7/2019 Stability of Nonlinear Systems

    27/34

    Example

    Continuous Biochemical Reactor

    Fresh Media Feed

    (substrates)

    Exit Gas Flow

    Agitator

    Exit Liquid Flow

    (cells & products)

    27

  • 8/7/2019 Stability of Nonlinear Systems

    28/34

    Cell Growth Modeling

    Specific growth rate

    Yield coefficients

    Biomass/substrate: YX/S = -(X/(S

    Product/substrate: YP/S = -(P/(S

    Product/biomass: YP/X =(P/(X

    Assumed to be constant

    Substrate limited growth

    S = concentration of rate limiting substrate

    Ks = saturation constant

    Qm = maximum specific growth rate (achieved when S >> Ks)

    (g/L)ionconcentratbiomass1

    !! Xdt

    dX

    XQ

    SK

    S

    SS

    m

    !

    Q

    Q )(

    28

  • 8/7/2019 Stability of Nonlinear Systems

    29/34

    Continuous Bioreactor ModelAssumptions

    Sterile feed

    Constant volume

    Perfect mixing

    Constant temperature &pH

    Single rate limitingnutrient

    Constant yields

    Negligible cell death

    Product formation rates Empirically related to specific growth rate

    Growth associated products: q =YP/XQ

    Nongrowth associated products: q =F

    Mixed growth associated products: q =YP/XQF29

  • 8/7/2019 Stability of Nonlinear Systems

    30/34

    Mass Balance Equations

    Cell mass

    VR = reactor volume

    F= volumetric flow rate

    D =F/VR = dilution rate Product

    Substrate

    S0 = feed concentration of rate limiting substrate

    XDXdt

    dXXVFX

    dt

    dXV

    RRQQ !!

    qXDPdt

    dPqXVFP

    dt

    dPV RR !!

    XY

    SSDdt

    dSXV

    YFSFS

    dt

    dSV

    SX

    R

    SX

    R QQ/

    0

    /

    0

    1)(

    1!!

    30

  • 8/7/2019 Stability of Nonlinear Systems

    31/34

    Steady-State Solutions Simplified model equations

    Steady-state equations

    Two steady-state points

    ),()(1

    )(

    )(),()(

    2

    /

    0

    1

    SXfXSY

    SSDdt

    dS

    SK

    SSSXfXSDX

    dt

    dX

    SX

    S

    m

    !!

    !!!

    Q

    QQQ

    0)(1

    )(

    )(0)(

    /

    0 !

    !!

    XSY

    SSD

    S

    SSXSXD

    SX

    S

    Q

    QQQ

    0:Washout

    )()(:Trivial-Non

    0

    0/

    !!

    !

    !!

    XSS

    SSYXD

    DSDS

    SX

    S

    QQ

    31

  • 8/7/2019 Stability of Nonlinear Systems

    32/34

    Model Linearization Biomass concentration equation

    Substrate concentration equation

    Linear model structure:

    ? A

    S

    SK

    SX

    SK

    XXDS

    SSS

    fXX

    X

    fSXf

    dt

    Xd

    S

    m

    S

    m

    SXSX

    d

    -

    d!

    -

    x

    x

    -

    x

    x$

    d

    2

    ,

    1

    ,

    1

    1

    zero

    ),(

    QQQ

    SD

    S

    SX

    S

    X

    YX

    S

    S

    Y

    SSS

    fXX

    X

    fSXf

    dt

    Sd

    S

    S

    SXS

    SX

    SXSX

    d

    -

    d

    !

    -

    x

    x

    -

    x

    x$

    d

    2

    //

    ,

    2

    ,

    2

    2

    11

    zero

    ),(

    QQQ

    SaXadt

    Sd

    SaXadt

    Xd

    dd!d

    dd!d

    2221

    1211

    32

  • 8/7/2019 Stability of Nonlinear Systems

    33/34

    Non-Trivial Steady State Parameter values

    KS = 1.2 g/L, Qm=0.48 h-1, Y

    X/S =0.4 g/g

    D =0.15 h-1, S0 = 20 g/L

    Steady-state concentrations

    Linear model coefficients (units h-1)

    529.3

    1375.0

    1

    472.10

    2

    /

    22

    /

    21

    21211

    !

    !!

    !

    !

    !!

    D

    S

    SX

    S

    X

    Ya

    S

    S

    Ya

    S

    SX

    S

    Xaa

    S

    S

    SXS

    SX

    S

    S

    QQQ

    QQ

    g/78.7)(g/545.0 0/ !!!! SSYXD

    DK

    S SXm

    S

    Q

    33

  • 8/7/2019 Stability of Nonlinear Systems

    34/34

    Stability Analysis Matrix representation

    Eigenvalues (units h-1)

    Conclusion The system is asymptotically stable

    Axxdt

    dx

    S

    Xx !

    -

    !

    -

    dd

    !529.3375.0

    472.10

    365.3164.0529.3375.0

    472.111

    !!

    ! PP

    P

    PPIA

    34