Scribd Logo
Upload

Welcome to Scribd!

  • 30 views

    Riccati

    Uploaded by

    Tushar Sanwarey
    1) The document frames the proof of the Riccati equation, which arises in control problems involving linear systems. 2) It derives the optimal control signal that minimizes a quadratic performance measure using the Hamilton-Jacobi-Bellman equation and properties of positive definite matrices. 3) Substituting the optimal control signal back into the HJB equation results in the famous Riccati equation, which is a nonlinear ordinary differential equation that can be used for regulator design problems.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd

    Riccati

    Uploaded by

    Tushar Sanwarey
    30 views3 pages
    1) The document frames the proof of the Riccati equation, which arises in control problems involving linear systems. 2) It derives the optimal control signal that minimizes a quadratic performance measure using the Hamilton-Jacobi-Bellman equation and properties of positive definite matrices. 3) Substituting the optimal control signal back into the HJB equation results in the famous Riccati equation, which is a nonlinear ordinary differential equation that can be used for regulator design problems.

    Copyright

    © © All Rights Reserved

    Available Formats

    DOCX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    1) The document frames the proof of the Riccati equation, which arises in control problems involving linear systems. 2) It derives the optimal control signal that minimizes a quadratic performance measure using the Hamilton-Jacobi-Bellman equation and properties of positive definite matrices. 3) Substituting the optimal control signal back into the HJB equation results in the famous Riccati equation, which is a nonlinear ordinary differential equation that can be used for regulator design problems.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    30 views3 pages

    Riccati

    Uploaded by

    Tushar Sanwarey
    1) The document frames the proof of the Riccati equation, which arises in control problems involving linear systems. 2) It derives the optimal control signal that minimizes a quadratic performance measure using the Hamilton-Jacobi-Bellman equation and properties of positive definite matrices. 3) Substituting the optimal control signal back into the HJB equation results in the famous Riccati equation, which is a nonlinear ordinary differential equation that can be used for regulator design problems.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    You are on page 1of 3

    Here, we frame the proof of the Riccati equation.

    Consider the HJB


    T
    ¿
    J t (t , x t )+min ( g(t , x t , ut )+J x (t , x t ) ẋ t )=0
    ¿

    ut

    For the nonlinear system described via nonlinear ordinary differential equation,

    ẋ t =a(t , x t ,u t )

    For its linear counterpart

    ẋ t =A t x t + Bt u t ,

    where ut is a ‘vector control signal’. We wish to design the ‘vector optimal control signal’.
    That results in the optimally controlled state ẋ t . Thus, exploit equation (1) and choose the
    performance measure and optimal performance measure using the quadratic function. Note
    that the properties of the performance measure agree with the properties of the norm of the
    state vector and adopting the property of the positive definite symmetric property. Thus,
    ¿ T T T
    J (t , x t )=xt K t xt . g(t , x t , ut )=x t Qt x t +u t Rt x t

    Thus
    .
    ¿ T
    J ¿t (t , x t )=xTt K t xt J x (t , x t )=x t K t x t =2 K t x t

    Substitute the above in the first equation


    .
    min ( xTt Qt x t +uTt R t ut +2 x t K t ( A t x t + Bt u t ))=0
    T
    x Tt K t x t + ut

    (2)

    To minimum the second term of the right-hand side, take the first-order partial derivative

    with respect to the control signal, we find an expression for the optimal control signal. Set

    ∂ ( x T Q x +uT R u +2 x K ( A x +B u ))=0
    ∂ut t t t t t t t
    T t t t t t

    As a result of this, we have

    2 R t ut +2 B T K t xt =0
    t

    Thus
    ut∗¿−R−1
    t B T K t xt
    t

    (3)

    This is the optimal control signal. This has ability to minimize the performance measure.

    This tells the feedback design. Now, we pose a question how to design the feedback

    parameter. This expression suggests a negative state feedback generates the optimal control

    signal. Now, the next question is to a regulator design? Equation (2) after combining with

    equation (3), we have


    . T
    x Tt K t x t + x Tt Q t x t +u¿t Rt u¿t +2 x K t ( At x t + Bt ut )=0 ¿
    tT

    The above further can be alternatively expressed as


    . T
    x Tt K t x t + x Tt Q t x t +u¿t Rt u¿t +2 x K t ( At x t + Bt ut ) ¿
    tT

    . T T
    =x Tt K t x t + x Tt Q t x t +u ¿t R t u¿t + x T K t A t x t + x T K t B t ut + x T ATt K t x t +u¿t BTt K t x t = 0
    ¿
    t t t

    (4)

    Combine (3) and (4), we have


    . T T
    x Tt K t x t + x Tt Q t x t +u¿t Rt u¿t + x K t At xt+ x K t B t ut + x ¿ A Tt K t x t + u¿t B Tt K t x t
    tT tT tT

    .
    −1 −1 −1 T T −1 T
    = x Tt K t x t + x Tt Q t x t + x Tt K Bt Rt Rt Rt B t K t xt + x t K t A t x t − x t K t Bt R t B t K t xt + x t At K t x t − x t K t Bt Rt B t K t x t
    T T T T T

    tt t

    =0.

    The above can rearranged as


    .
    −1 −1 T −1 T
    x Tt ( K t +Q t + K B t R t Bt K t + K t A t − K t Bt Rt B t K t + A t K t − K t Bt R t Bt K t ) x t = 0.
    T T

    tt t

    For nonzero state vector , a nontrivial condition is


    .
    −1 T
    K t +Q t + K t At + ATt K t − K Bt Rt B t K t =0 .
    tt

    The above is a celebrated Riccati equation. It has structure of the second-order nonlinear
    ordinary differential equation. Note that the matrices Qt K t ,Rt are the positive definite
    symmetric matrices. This is also known as the LQR equation.
    Note that the Riccati equation arises in the Kalman filtering, stability of linear time-delay system,
    control of the linear system driven by the input signal.

    Understand the following.

    Riccati equation arises in the Kalman filtering?

    stability of linear time-delay system?

    control of the linear system driven by the input signal (explained from the HJB).

    Reference: S Bittai, J. Laub and Jan C.Willems (Ed.), the Riccati equation, Springerr-
    Verlag, New York, 1991.

    You might also like