Scribd Logo
Upload

Welcome to Scribd!

  • 93 views

    Exercise 8: LQG

    Uploaded by

    Sreeja Sunder
    This document describes exercises for a course on state-space control systems. The first exercise asks to derive matrices for the closed loop system of an LQG controller. The second asks to write the recursive Kalman filter equations. The remaining exercises involve simulating an RLC circuit system and designing an LQG controller using Luenberger observer, Kalman filter, and LQR feedback to stabilize the system with and without voltage measurement.

    Copyright:

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

    Exercise 8: LQG

    Uploaded by

    Sreeja Sunder
    93 views2 pages
    This document describes exercises for a course on state-space control systems. The first exercise asks to derive matrices for the closed loop system of an LQG controller. The second asks to write the recursive Kalman filter equations. The remaining exercises involve simulating an RLC circuit system and designing an LQG controller using Luenberger observer, Kalman filter, and LQR feedback to stabilize the system with and without voltage measurement.

    Original Title

    exercise08

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document describes exercises for a course on state-space control systems. The first exercise asks to derive matrices for the closed loop system of an LQG controller. The second asks to write the recursive Kalman filter equations. The remaining exercises involve simulating an RLC circuit system and designing an LQG controller using Luenberger observer, Kalman filter, and LQR feedback to stabilize the system with and without voltage measurement.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    93 views2 pages

    Exercise 8: LQG

    Uploaded by

    Sreeja Sunder
    This document describes exercises for a course on state-space control systems. The first exercise asks to derive matrices for the closed loop system of an LQG controller. The second asks to write the recursive Kalman filter equations. The remaining exercises involve simulating an RLC circuit system and designing an LQG controller using Luenberger observer, Kalman filter, and LQR feedback to stabilize the system with and without voltage measurement.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    Exercises for Course on State-Space Control Systems (SSC)

    Albert-Ludwigs-Universität Freiburg – Summer Term 2019


    Exercise 8: LQG
    Prof. Dr. Moritz Diehl, Dr. Dang Doan, Benjamin Stickan, Katrin Baumgärtner

    Theoretical Exercises
    1. Consider a dynamic system of the form

    ẋ = Ax + Bu + w, E(w(s)w> (t)) = Qw δ(t − s)


    y = Cx + v, E(v(s)v > (t)) = Rv δ(t − s),

    (a) If the system is LQG controlled, the closed loop system can be described by
         
    ẋ x w
    ˙x̂ = Ã x̂ + B̃ v .

    Derive matrices à and B̃.


    (b) How do the poles of the closed loop system change if we describe it by
         
    ẋ x w
    = Ãe + B̃e
    ė e v

    with e := x − x̂.

    2. Write down the recursive Kalman filter equations for continuous time and explain them.

    MATLAB/Simulink: LC resonant circuit


    The electrical circuit sketched below shows a resonant LC circuit.

    iL (t) L

    v(t) C vC (t)

    The system can be described in state-space representation by

    ẋ = Ax + Bu,
    y = Cx,

    with
    −1
      1  
    0 L L
    1 0
    A= 1 ,B = and C = . (1)
    C
    0 0 0 1
     |
    The state vector is defined as x := iL vC and the input as u := v.

    The initial state of the simulation used in this exercise is x0 = [0, 5]. Our aim is to design an LQG controller
    that stabilizes the system.

    1. Open the Simulink template ex08_sim.slx and implement a continuous Luenberger observer with
    poles [−1, −1]. Use the ex08_init.m file to define the observer gain L. Compare x and x̂luen .

    1
    2. Add process disturbance w and measurement noise v, such that the system becomes stochastic:

    ẋ = Ax + Bu + w, E(w(s)w> (t)) = Qw δ(t − s)


    y = Cx + v, E(v(s)v > (t)) = Rv δ(t − s),

    with
           
    rw1 0 0.001 0 rv1 0 1 0
    Qw = = , Rv = = .
    0 rw2 0 0.001 0 rv2 0 1

    Hint: Use the Band-Limited White Noise block with noise powers rw1 , rw2 , rv1 , rv2 and a
    sampling time t = 10 µs. Also use different integer seeds for all random signals.

    3. Implement the continuous recursive Kalman filter for the stochastic system. Use
       
    0 1 1
    x̂0 = and P0 =
    5 1 1

    as the initial states.

    4. Observe the trajectory of the error covariance matrix of the Kalman filter. How can we simplify the
    filter structure?

    5. Implement an LQR controller which feeds back

    (a) the noisy measured states y = x + v


    (b) the estimated state x̂ from the Kalman filter

    Use weighting matrices

     
    0.1 0
    Q= and R = 0.1
    0 1

    6. Now we will remove the voltage measurement of our system. Copy the plant and the Kalman filter in
    a new subsystem and replace the output matrix C by an appropriate matrix C2 . You will also have to
    adapt the measurement noise in your plant as well as in the Kalman filter.

    7. Change the initial state of the Kalman filter to


     
    0
    x̂0 =
    0

    and explain the behavior of the LQG controlled closed loop system.

    You might also like