Scribd Logo
Upload

Welcome to Scribd!

  • 19 views

    Algorithm EKF

    Uploaded by

    ramiyamin
    The document describes a system identification model using an extended Kalman filter (EKF). It initializes state and covariance matrices and defines the system matrices A, C, and their equations. It then runs the EKF iteratively for 600 time steps, using the measurement data to correct the state estimate. The Jacobian of the A matrix is also computed to linearize the model for the EKF updates.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd

    Algorithm EKF

    Uploaded by

    ramiyamin
    19 views5 pages
    The document describes a system identification model using an extended Kalman filter (EKF). It initializes state and covariance matrices and defines the system matrices A, C, and their equations. It then runs the EKF iteratively for 600 time steps, using the measurement data to correct the state estimate. The Jacobian of the A matrix is also computed to linearize the model for the EKF updates.

    Original Description:

    ekf

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document describes a system identification model using an extended Kalman filter (EKF). It initializes state and covariance matrices and defines the system matrices A, C, and their equations. It then runs the EKF iteratively for 600 time steps, using the measurement data to correct the state estimate. The Jacobian of the A matrix is also computed to linearize the model for the EKF updates.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    19 views5 pages

    Algorithm EKF

    Uploaded by

    ramiyamin
    The document describes a system identification model using an extended Kalman filter (EKF). It initializes state and covariance matrices and defines the system matrices A, C, and their equations. It then runs the EKF iteratively for 600 time steps, using the measurement data to correct the state estimate. The Jacobian of the A matrix is also computed to linearize the model for the EKF updates.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 5

    Begin

    R0
    U_OC
    i
    Y

    =
    =
    =
    =

    System
    0.0224;
    12.7;
    -3.8;
    Load (mydata);

    parameters:
    (considered as constant)
    (could be variable)
    (Charge current)
    (output measurement data)

    Initialization:
    X0= [0.10 0.2 59 1000 0.005 10 0.0001]'

    (1)

    P0= [10 0 0 0 0 0 0
    0 10 0 0 0 0 0
    0 0 10 0 0 0 0
    0 0 0 10 0 0 0
    0 0 0 0 10 0 0
    0 0 0 0 0 10 0
    0 0 0 0 0 0 10]
    P_Predected = P0 ;

    P_estimated = P0 ;

    The systems
    A = [a11 0
    0
    0
    a22 0
    a31 0
    a33
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0

    Matrices:
    a14 a15 0
    0
    0
    a26
    a34 a35 a36
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0

    0
    a27
    a37
    0
    0
    0
    0]; (2)

    C = [0

    0];

    Simulation time: T = 600;


    Sampling period dt = 0.1;

    EKF iteration k=1:


    Prediction
    P_Predected
    = A* P_estimated *A';
    K (:, 1)
    =(P_Predected*c')/(c* P_Predected *c');
    X_Predected (:, 1) = A*X0;
    Update
    X_estimated(:,1)=X_Predected(:,1)+K(:,1)*(Y -Y_predected);

    (
    (

    )
    )

    j = 2

    j = j + 1

    False
    2 j 600

    True

    EKF _ Compute the Jacobian of A matrix:


    A = [a11 0
    0
    a14 a15 0
    0
    0
    a22 0
    0
    0
    a26 a27
    a31 0
    a33 a34 a35 a36 a37
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0
    0];
    (3)
    P_Predected
    K(:,j)
    X_Predected(:,j)
    Y_estimated

    =
    =
    =
    =

    Prediction
    A* P_estimated *A';
    (P_Predected*C')/(C*P_Predected *C');
    A*X_estimated(:,j-1);
    C* X_Predected(:,j)

    Update
    X_estimated(:,j)=X_Predected(:,j)+K(:,j)*(Y(j)-Y_predected);

    (
    (

    )
    )

    Plote results
    END

    With:
    (1)
    x= [Upa Upc UL 1/Rpa 1/Cpa 1/Rpc 1/Cpa]T

    (2)
    a11=-X0(4)*X0(5);
    a14=-X0(1)*X0(5);
    a15=-X0(1)*X0(4)+i;
    a22=-X0(6)*X0(7);
    a26=-X0(2)*X0(7);
    a27=-X0(2)*X0(6)+i;
    a31=X0(4)*X0(5)-X0(6)*X0(7);
    a33=-X0(6)*X0(7);
    a34=X0(5)*X0(1);
    a35=X0(4)*X0(1)-i;
    a36=-X0(3)*X0(7)-X0(1)*X0(7)-R0*X0(7)*i+U_OC*X0(7);
    a37=-X0(6)*X0(3)-X0(6)*X0(1)-R0*X0(6)*i-i+U_OC*X0(6);
    (3)
    a11= -X_Predected(4,j-1)* X_Predected(5,j-1);
    a14= -X_Predected(1,j-1)* X_Predected(5,j-1);
    a15= -X_Predected(1,j-1)* X_Predected(4,j-1)+i;
    a22= -X_Predected(6,j-1)* X_Predected(7,j-1);
    a26= -X_Predected(2,j-1)* X_Predected(7,j-1);
    a27= -X_Predected(2,j-1)* X_Predected(6,j-1)+i;
    a31= X_Predected (4,j-1)* X_Predected(5,j-1)X_Predected(6,j-1)*X_Predected(7,j-1);
    a33=-X_Predected(6,j-1)* X_Predected(7,j-1);
    a34= X_Predected(5,j-1)* X_Predected(1,j-1);
    a35= X_Predected(4,j-1)* X_Predected(1,j-1)-i;
    a36= -X_Predected(3,j-1)*X_Predected(7,j-1)X_Predected(1,j-1)*X_Predected(7,j-1)-R0*
    X_Predected(7,j-1)*i+U_OC*X_Predected(7,j-1);
    a37=- X_Predected(6)* X_Predected(3)- X_Predected(6)*
    X_Predected(1)-R0*(6)*i-i+U_OC* X_Predected(6);

    The models equations:

    Identification by EKF :
    Assume that:

    Transformation for the EKF:


    (

    )
    (

    The Jacobian of the A matrix:

    The C matrix :

    You might also like