State-Space Representations of Transfer Function Systems

Burak Demirel February 2, 2013

1 State-Space Representation in Canonical Forms

We here consider a system dened by y (n) + a1 y (n1) + + an1 y + an y = b0 u(n) + b1 u(n1) + + bn1 u + bn u , where u is the control input and y is the output. We can write this equation as b0 sn + b1 sn1 + + bn1 s + bn Y (s) = n . U (s) s + a1 sn1 + + an1 s + an (2) (1)

Later, we shall present state-space representation of the system dened by (1) and (2) in controllable canonical form, observable canonical form, and diagonal canonical form.

1.1 Controllable Canonical Form

We consider the following state-space representation, being called a controllable canonical form, as x 1 0 1 0 ... 0 x1 0 x 0 1 ... 0 x2 0 2 0 . . . . . . . .. . . . . . . = + (3) . u . . . . . . . . x n1 0 0 0 ... 1 xn1 0 x n an an1 an2 . . . a1 xn 1 x1 x2 y = (bn an b0 ) (bn1 an1 b0 ) . . . (b1 a1 b0 ) . + b0 u (4) . . xn Note that the controllable canonical form is important in dicsussing the pole-placement approach to the control system design.

1.2 Observable Canonical Form

We consider the following state-space representation, being called an observable canonical form, as x1 x 1 0 0 . . . 0 an bn an b0 x 2 1 0 . . . 0 an1 x2 bn1 an1 b0 x bn2 an2 b0 (5) 3 = 0 1 . . . 0 an2 x3 + u . . . . . . . . . . . . . . ... . . . . . . . b a1 b0 1 xn x n 0 0 . . . 1 a1 x1 x2 (6) y = 0 0 0 . . . 1 x3 + b0 u . . . xn

1.3 Diagonal Canonical Form

We here consider the transfer function system given by (2). We have the case where the dominator polynomial involves only distinct roots. For the distinct root case, we can write (2) in the form of Y (s) b0 sn + b1 sn1 + + bn1 s + bn = U (s) (s + p1 )(s + p2 ) (s + pn ) c1 c2 cn = b0 + + + + . s + p1 s + p2 s + pn (7) (8)

The diagonal canonical form of the state-space representation of this system is given by x 1 p1 0 x1 1 x p x 2 2 2 1 . . . . .. (9) . = . + . u . . . x xn1 1 pn1 n1 x n 0 pn xn 1 x1 x2 . (10) y = c1 c2 . . . cn1 cn . + b0 u . xn1 xn

2 Numerical Examples
Example 1: Obtain the transfer function of the system dened by the following statespace equations: x 1 0 1 0 x1 0 x 2 = 0 0 1 x2 + 0 u , (11) x 3 6 11 6 x3 1 x1 (12) y = 1 0 0 x2 . x3 Solution: From (11) and (12), we determine the following parameters: b0 = 0, b1 = 0, b2 = 0, b3 = 1, a1 = 6, a2 = 11, a3 = 6. Thus, the resulting transfer function is G(s) = 1 Y (s) = 3 . U (s) s + 6s2 + 11s + 6

Example 2: Find the state-space representation of the following transfer function system (13) in the diagonal canonical form. G(s) = s2 2s + 3 . + 5s + 6 (13)

Solution: Partial fraction expansion of (13) is 2s + 3 A B = + . s2 + 5 s + 6 s+2 s+3 Hence, we get A = 1 and B = 3. We now have two distinct poles. For this, we can write the transfer function (13) in the following form: x 1 2 0 = x 2 0 3 y = 1 3 x1 x2 x1 1 + u x2 1 (14) (15)

Example 3: Obtain the state-space representation of the transfer function system (16) in the controllable canonical form. G(s) = s2 + 3s + 3 s2 + 2s + 1 (16)

Solution: From the transfer function (16), we obtain the following parameters: b0 = 1, b1 = 3, b2 = 3, a1 = 2, and a2 = 1. The resulting state-space model in controllable canonical form is obtained as x 1 0 1 = x 2 1 2 y= 2 1 x1 0 + u, x2 1

(17) (18)

x1 +u. x2

Example 4: Consider the following state equations: x 1 = x2 (t) , x 2 = x3 (t) , x 3 = 6x1 (t) 11x2 (t) 6x3 (t) + 6u(t) , y = x1 (t) , and determine the controllable canonical form. Solution: Using the state equations (19), (20), (21), and (22), we write the following high order dierential equation: d3 d2 d y ( t ) + 6 y (t) + 11 y (t) + 6y (t) = 6u(t) . 3 2 dt dt dt The state variables x(t) = y , x2 (t) = y , and x3 = y . Hence, we get x 1 =x2 (t) , x 2 =x3 (t) , x 3 = d3 y (t) = 6 y (t) 11y (t) 6y (t) + 6u(t) dt3 = 6x3 (t) 11x2 (t) 6x1 (t) + 6u(t) . (19) (20) (21) (22)

In matrix form, we have 0 0 1 0 0 1 x(t) + 0 u(t) , x (t) = 0 6 6 11 6 y (t) = 1 0 0 x(t) . Example 5: Consider the following state equations x 1 (t) = x1 (t) + x3 (t) + 4u(t) , x 2 (t) = 3x1 (t) + 2u(t) , x 3 (t) = 5x1 (t) + x2 (t) + u(t) , y (t) =x1 (t) , 4 (23) (24) (25) (26)

and determine the observable canonical form. Solution: Using the state equations (23), (24), (25), and (26), we write the following high order dierential equation: d3 d2 d d2 d y ( t ) + y ( t ) + 5 y ( t ) + 3 y ( t ) = 4 u(t) + u(t) + 2u(t) . 3 2 2 dt dt dt dt dt We introduce x1 (t) = y (t) in the equation, and collect all terms without dierentiation on the right hand side, we get d3 d2 d d2 d x ( t ) + x ( t ) + x ( t ) u(t) u(t) = 3x1 (t) + 2u(t) , 1 1 1 3 2 2 dt dt dt dt dt i.e., d d d d2 x1 (t) + x1 (t) + x1 (t) u(t) u(t) = 3x1 (t) + 2u(t) . 2 dt dt dt dt Now introduce the expression within the paranthesis as a new state variable x2 (t) = i.e., x 2 (t) = 3x1 (t) + 2u(t) . Repeating this precedure yields d d x1 (t) + x1 (t) 4u(t) = x2 (t) 5x1 (t) + u(t) , dt dt and we introduce x3 (t) = i.e., x 1 (t) = x3 (t) x1 (t) + 4u(t) . From (27), (28), and (29), we dene the 1 0 x (t) = 3 0 5 1 state-space form of 1 4 0 x(t) + 2 u(t) , 0 1 (29) d x1 (t) + x1 (t) 4u(t) , dt (28) (27) d2 d d x1 (t) + x1 (t) + 5x1 (t) 4 u(t) u(t) , dt2 dt dt

y (t) = 1 0 0 x(t) .

3 References
1. Katsuhiko Ogata, Modern Control Engineering, 4th Ed., Prentice Hall Inc., New Jersey, 2002. 2. KTH Reglerteknik, Reglerteknik Allm an Kurs, Del 2, 2007.