Skip to main content

Controllable, observable and diagonal forms

Short answer

Controllable form: G=[0 1; −0.6 −1], H=[0;1], C=[1 1]. Observable form: G=[0 −0.6; 1 −1], H=[1;1], C=[0 1]. Diagonal form uses poles −0.5±j0.5916 and residues 0.5∓j0.4226, giving diagonal G with conjugate entries and H=[1;1], C=[0.5−j0.4226, 0.5+j0.4226].

Consider the system defined by Y(z)/U(z)=(z+1)/(z²+z+0.6). Obtain state-space representation for this system in the following three different forms: (i) controllable canonical form (ii) observable canonical form (iii) diagonal canonical form.

The three forms all realise

Y(z)U(z)=z+1z2+z+0.6\frac{Y(z)}{U(z)} = \frac{z+1}{z^2+z+0.6}

with a1=1a_1 = 1, a2=0.6a_2 = 0.6, b0=0b_0 = 0, b1=1b_1 = 1, b2=1b_2 = 1.

(i) Controllable canonical form

Let Q(z)U(z)=1z2+z+0.6\dfrac{Q(z)}{U(z)} = \dfrac{1}{z^2+z+0.6}, so Y=(z+1)QY = (z+1)Q. With x1(k)=q(k)x_1(k) = q(k) and x2(k)=q(k+1)x_2(k) = q(k+1):

x1(k+1)=x2(k)x2(k+1)=−0.6x1(k)−x2(k)+u(k)y(k)=x1(k)+x2(k)\begin{aligned} x_1(k+1) &= x_2(k)\\ x_2(k+1) &= -0.6x_1(k) - x_2(k) + u(k)\\ y(k) &= x_1(k) + x_2(k) \end{aligned} x(k+1)=[01−0.6−1]x(k)+[01]u(k),y(k)=[11]x(k)\mathbf{x}(k+1) = \begin{bmatrix} 0 & 1 \\ -0.6 & -1 \end{bmatrix}\mathbf{x}(k) + \begin{bmatrix} 0 \\ 1 \end{bmatrix}u(k),\qquad y(k) = \begin{bmatrix} 1 & 1 \end{bmatrix}\mathbf{x}(k)

(ii) Observable canonical form

Y=z−1[U−Y+z−1(U−0.6Y)]Y = z^{-1}\left[U - Y + z^{-1}(U - 0.6Y)\right]. Take x2=yx_2 = y and X1=z−1(U−0.6Y)X_1 = z^{-1}(U - 0.6Y):

x1(k+1)=−0.6x2(k)+u(k)x2(k+1)=x1(k)−x2(k)+u(k)y(k)=x2(k)\begin{aligned} x_1(k+1) &= -0.6x_2(k) + u(k)\\ x_2(k+1) &= x_1(k) - x_2(k) + u(k)\\ y(k) &= x_2(k) \end{aligned} x(k+1)=[0−0.61−1]x(k)+[11]u(k),y(k)=[01]x(k)\mathbf{x}(k+1) = \begin{bmatrix} 0 & -0.6 \\ 1 & -1 \end{bmatrix}\mathbf{x}(k) + \begin{bmatrix} 1 \\ 1 \end{bmatrix}u(k),\qquad y(k) = \begin{bmatrix} 0 & 1 \end{bmatrix}\mathbf{x}(k)

(iii) Diagonal canonical form

Poles:

z2+z+0.6=0  ⇒  p1,2=−1±1−2.42=−0.5±j0.5916z^2 + z + 0.6 = 0 \;\Rightarrow\; p_{1,2} = \frac{-1 \pm \sqrt{1-2.4}}{2} = -0.5 \pm j0.5916

They are distinct, but complex. Residues:

c1=[z+1z−p2]z=p1=0.5+j0.5916j1.1832=0.5−j0.4226c2=cˉ1=0.5+j0.4226\begin{aligned} c_1 &= \left[\frac{z+1}{z-p_2}\right]_{z=p_1} = \frac{0.5 + j0.5916}{j1.1832} = 0.5 - j0.4226\\ c_2 &= \bar{c}_1 = 0.5 + j0.4226 \end{aligned} Y(z)U(z)=0.5−j0.4226z+0.5−j0.5916+0.5+j0.4226z+0.5+j0.5916\frac{Y(z)}{U(z)} = \frac{0.5 - j0.4226}{z + 0.5 - j0.5916} + \frac{0.5 + j0.4226}{z + 0.5 + j0.5916}

With Xi=U/(z−pi)X_i = U/(z - p_i):

x(k+1)=[−0.5+j0.591600−0.5−j0.5916]x(k)+[11]u(k)\mathbf{x}(k+1) = \begin{bmatrix} -0.5 + j0.5916 & 0 \\ 0 & -0.5 - j0.5916 \end{bmatrix}\mathbf{x}(k) + \begin{bmatrix} 1 \\ 1 \end{bmatrix}u(k) y(k)=[0.5−j0.42260.5+j0.4226]x(k)y(k) = \begin{bmatrix} 0.5 - j0.4226 & 0.5 + j0.4226 \end{bmatrix}\mathbf{x}(k)

The states are complex conjugates of each other, but the output y(k)y(k) is real. Exact values: p1,2=−12±j3510p_{1,2} = -\tfrac12 \pm j\tfrac{\sqrt{35}}{10} and c1,2=12∓j3514c_{1,2} = \tfrac12 \mp j\tfrac{\sqrt{35}}{14}.

Check

  • c1+c2=1c_1 + c_2 = 1, the coefficient of z in the numerator.
  • −(c1p2+c2p1)=1-(c_1p_2 + c_2p_1) = 1, the constant term.

All three forms give z+1z2+z+0.6\dfrac{z+1}{z^2+z+0.6}. The poles have ∣p∣=0.6=0.775<1|p| = \sqrt{0.6} = 0.775 < 1, so the system is stable.

✓ Answer checked by Claude Opus 5.5 (high effort)

All Digital Control System chapter 5 questions →