lqr

Linear-quadratic regulator for a continuous-time model.

Syntax

[G, X, L] = lqr(SYS, Q, R)

[G, X, L] = lqr(SYS, Q, R, N)

[G, X, L] = lqr(A, B, Q, R)

[G, X, L] = lqr(A, B, Q, R, N)

[G, X, L] = lqr(A, B, Q, R, N, E)

Inputs

SYS
Continuous or discrete-time LTI model.
Q
The state weighting matrix (n x n), which is symmetric, positive semi-definite.
R
The input weighting matrix (p x p), which is symmetric, positive definite.
N
The state/input cross product weighting matrix, such that Q - N*inv(R)*N' is positive semi-definite.
A
The state matrix (n x n).
B
The input matrix (n x p).
E
The descriptor matrix (n x n).

Outputs

G
The feedback gain matrix.
X
The solution of the Continuous Algebraic Riccati Equation.
L
The closed-loop pole locations, the eigenvalues of the matrix A-BK.

Examples

LQR design:
A=[0.8 0;0.8 1];
B=[0.6;0.6];
C=[0 1];
Q=C'*C;
R=1;
[G, X, L] = lqr(A, B, Q, R)
G = [Matrix] 1 x 2
-1.77636e-15  6.16228e+00
X = [Matrix] 2 x 2
 8.21637  -8.21637
-8.21637  18.48683
L = [Matrix] 2 x 1
-1.26491
-0.63246
LQR design from a state-space model:
   A = [-0.313 56.7 0; -0.0139 -0.426 0; 0 56.7 0];
    B = [0.232; 0.0203; 0];
    C = [0 0 1];
    D = [0];
    sys = ss(A, B, C, D);
    p = 2;
    Q = p * transpose(C) * C;
    R = 1;
    [G, X, L] = lqr(sys, Q, R)
G = [Matrix] 1 x 3
-0.50335  52.86451  1.41421
X = [Matrix] 3 x 3
  1.41413   -40.95711  -1.66409
-40.95711  3072.24438  88.68382
 -1.66409    88.68382   3.64894
L = [Matrix] 3 x 1
-0.78084 + 1.12564i
-0.78084 - 1.12564i
-0.13369 + 0.00000i

Comments

[G, X, L] = lqr(SYS, Q, R, N) calculates the optimal gain matrix G.
Equations:
          .
          x = A x + B u,   x(0) = x0
          
                  inf
          J(x0) = INTEGRAL (x' Q x  +  u' R u  +  2 x' N u)  dt
                   0
          
          L = eig (A - B*G)

Comments

The function minimizes a quadratic cost function for a linear state-space system model.