Chapter 3.

Modeling in the Time Domain

3.1 Introduction
State space (modern or time-domain) approach:
• Nonlinear system (backlash, saturation, dead zone, …)
• System with nonzero initial conditions
• Time-varying system: missiles with varying fuel levels, lift in aircraft flying
• MIMO (multiple-input, multiple-output) system
• System simulation with digital computer

-1-
 3.2 Some Observations

 di 1
 L dt + Ri + C ∫ i dt =v(t )

 dq di d 2 q
 =with i (t ) =and = 2
, ∫ idt q
 dt dt dt
 d 2q dq 1
 → L 2
+R + q= v(t )
vL (t ) q  dt dt C
di 1
L + Ri + ∫ i dt =
v(t )
dt C
di 1  dq
L − q (t ) − Ri (t ) + v(t )
= =i
dt C  dt
1 → →
vL (t ) = − q (t ) − Ri (t ) + v(t )
C di 1 R 1  di 1 R 1
=
− q − i + v(t ) =
− q − i + v(t )
dt LC L L  dt LC L L

⇒ Output equation: linear combination of q(t), i(t), and input, v(t)

⇒ State equation: x = Ax + Bu ⇒ State-space representation  0 1 

A =  −1 − R 

Output equation: =
y Cx + Du  dq   LC L 
 dt   0 1  0
  q  0
State x:  dq 
=  
 −1 − R    +  1  v(t ) B= 1
 di  i     
 dt   q′ q
   dt   LC L  L  L 
=x =   = , x    −1 
 di   i′  i  −1 q 
=C 
C
− R

 dt  vL= (t ) [ − R ]   + 1 ⋅ v(t )
C i  D =1

-2-
 Ex: Determine linearly independent or linearly dependent.
- Two vectors:
= x1 (1,1),
= x2 (2, 2)

3.3 The General State-Space Representation ⇓

Two vectors:
= x1 (1,1),
= x2 (2, 2)
S =K1x1 + K 2 x2 =K1 (1,1) + K 2 (2, 2)
( K1 2 K 2 , K1 + 2 K 2 )
=+
if K1 =
−2, K 2 =
1 then S =
0
Since K1 and K 2 are not zero,
Definitions: the two vectors are linearly dependent.

1. Linear combination:
S = K n xn + K n −1 xn −1 +  + K1 x1

2. Linear independence:
S=0 only if every Ki=0 and no xi=0.

3. System variables: Any variables that responds to an input or initial conditions.

q, i 4. State variables: Linearly independent system variables. Determine the value
q 
of all system variables for all t ≥ t0 .
x= 
i 
x 
5. State vector: Elements of a vector are the state variables.
x =  1
 x2 
6. State space: The n-dim. space whose axes are the state variables.
7. State equations: A set of n simultaneous, first-order differential equations.
8. Output equation: the output variables represented with a linear combination
of the state variables and the inputs.
x Ax + Bu
=
y Cx + Du
= -3-
 • Linear independence: S=0 only if every Ki=0 and no xi=0.
Example 1:
Consider the set of vectors {(1,2,0,0),(0,4,4,0),(2,0,3,0)} on the ℝ4 vector space.
Determine if this set is linearly independent or linearly dependent.

K1 (1, 2,0,0) + K 2 (0, 4, 4,0) + K 3 (2,0,3,0) =

0
2 K1 + 4 K 2 = 0  1 0 2  K1   0 
    
( K1 , 2 K1 ,0,0) + (0, 4 K 2 , 4 K 2 ,0) + (2 K 3 ,0,3K 3 ,0) = 0 → →  2 4 0  K 2  =
 0  --- (A)
4 K 2 + 3K 3 = 0  0 4 3  K   0 
( K1 + 2 K 3 , 2 K1 + 4 K 2 , 4 K 2 + 3K 3 , 0) = (0, 0, 0, 0)   3   

Ax = b → x = A−1b if A ≠ singular matrix (det( A) ≠ 0)

1 0 2 >> A=[1 0 2; 2 4 0; 0 4 3];

  >> det(A)
det  2 4 0  = 28
 0 4 3 ans = 28
 

A is invertible (det( A) ≠ 0). It implies that the equation (A) has only the trivial solution K1 =
K2 = 0.
K3 =
→ Therefore, this set of vectors is linearly independent.

Example 2:
 1 0   0 −1  1 0  
Consider the set of vectors  , ,   on the vector space M 22 .
 0 1   0 0   0 0   OR
K1 + K 3 =0  1 0 1  K1   0 
Determine if this set is linearly independent or linearly dependent.     
− K 2 = 0 →  0 −1 0  K 2  =  0 
K1 = 0  1 0 0  K   0 
0
K1 + K 3 =   3   
1 0  0 −1 1 0  K1 + K 3 −K2  1 0 1
K1   + K2   + K3  =0→  = 0 ⇒ −K2 = 0  
0 1 0 0  0 0  0 K1 
K1 = 0
→ det  0 −1 0  =
1 (≠ 0)
1 0 0
 
→ K1 = K 2 = K 3 = 0 → Therefore, this set of vectors is linearly independent.
-4-
3.4 Applying the State-Space Representation

Example 3.1: Representing an electrical network x Ax + Bu

=
Find a state-space representation if the output is iR . =y Cx + Du

diL
vL = L
dt
dvC  dvC
iC = C
dt  C dt = iC

 L diL = v
 dt L

vC dv dvC v dvC 1 1
iL = iR + iC = +C C ( KCL) C − C + iL
= =
− vC + iL
R dt dt R dt RC C
⇒ ⇒
di di diL 1 1
v(t ) = vL + vC = L L + vC ( KVL) L L =−vC + v(t ) − vC + v
=
dt dt dt L L

 −1 1
 RC 0
 vC  C   vC   
=      + 1 v(t )
 iL   −1 i  
0   L   L 
⇒ 
 L 
VC 1 v 
iR
= = ( 0)  C 
R R  iL 
-5-
 Example 3.2: Representing an electrical network
Find the state and output equations if the output is y = [vR 2
iR2 ]T .
i 
x = L 
 vC 

di dv  diL 
dvC L,
L L v=
= C C iC  dt 
C
dt dt dt ⇒ x =  
 dvC 
State variables
= : x1 iL=
, x2 vC  
 dt 

vR2 =−vC + vL
 vL dvC
i
 R= + iL + C
 1 dt iR2= iC + 4vL

 dvC vR2
C = − 4 v =
vL − vC
− 4 v =
(1 − 4 R2 ) v − 1 v
 dt R2
L
R2
L
R2
L
R2
C
 State variables :
= , x2 vC
x1 iL=
(v=
L vC + vR2 )
diL  diL 
vL = L =
 iL 
x = , x

 dt 
 
dt 
 vC   dvC 
 
 dt 

 1 diL dvC  dVC L diL

= i ( L ) + i + C C (
 dt ) + ( )= i − iL
 R dt
L
dt R dt
  1

1
⇒ 
C dvC (1 − 4 R2 ) ( L diL ) − 1 v C ( dVC ) − L(1 − 4 R2 ) ( diL ) =1
− VC
= 
 dt R2 dt R2
C
 dt R2 dt R2

-6-
  diL 
 iL   dt 
=x =  , x    dVC L diL
 vC   dvC   C ( dt ) + R ( dt ) = i − iL
Example 3.2: (Continued) 
 dt 
 

1

 C ( dVC ) − L(1 − 4 R2 ) ( diL ) =1

− VC
 dt R2 dt R2


Ci − iL
↓
1 C
C − VC − VC − C (i − iL )  L   dVC   i −i 
diL R2 R2 + R2iL − VC − R2i C   
= =
L
R1
   dt  =  1 
dt L L(1 − 4 R2 ) L L ∆  L(1 − 4 R2 )   diL   − VC 
C C[ − − ] C −    R2 
R1 R2 R1  R2   dt 

L(1 − 4 R2 )
C −
R2

1
(1 − 4 R2 )iL + VC − (1 − 4 R2 )i
dVC R1  R 
= , where ∆ = − (1 − 4 R2 ) + 2 
dt C ∆  R1 

 vR2   −vC + vL 
The output equation is: ⇒=y =
 iR   iC + 4vL 
  2

 vR2   R2 / ∆ −(1 + 1 / ∆ )   iL   − R2 / ∆ 
⇒      +   i (t )
i
 R2   1 / ∆ (1 − 4 R1 ) / ( ∆R ) v
1  C   −1 / ∆ 

-7-
Example 3.3: Representing a translational mechanical system Stiffness k x(t)

Find the state equations. Damping D

dx(t )
dt
= Dv(t )

d 2 x(t )
Inertia M = M v(t )
dt

x Ax + Bu
=

• State variables:  x1 , v1

 x2 , v2
 dx
 1
= v1
 x1 = v1  dt
  dx
k D k  2
= v2
 M v = k ( x2 − x1 ) − Dv1 ⇒ v1 =
− x1 − v1 + x2  dt
 1 1 M1 M1 M1

 x2 = v2
 k k 1
 M 2 v2 =f − k ( x2 − x1 ) ⇒ v2 = x1 − x2 + f x Ax + Bu
=
 M 2 M 2 M 2 y Cx + Du
=

 x1   0 1 0 0  x1   0   x1 
        
⇒  v1   − K / M1 − D / M1 K / M1 0  v1   0 
+ f (t ),  v1 
x2 = (0 0 1 0)  
 x2   0 0 0 1  x2   0  x
 2
       v 
 K / M2 0 −K / M 2 0  v2  1 / M 2   2
 v2 

-8-
 3.5 Converting a Transfer Function to State Space
 x1 
• Consider differential equation:  
x
xn x2 x1 x= 2
  
xn n
d y n −1
d y dy  
+ a n −1 +  + a1 + a0 y =
b0 u  xn 
dt n dt n −1 dt x Ax + Bu
=

dy d2y d n −1 y
Choose state variables ⇒ = , x2
x1 y= , x3
= , =
, xn
dt dt 2 dt n −1

dy d2y d3y dny

=
Differentiating eq.: ⇒ x1 = , x2 = , x3 , =
, xn
dt dt 2 dt 3 dt n

State eq. ⇒ x1 =x2 , x2 =x3 ,  , xn −1 =xn , xn =− a0 x1 − a1 x2 −  − an −1 xn + b0u

 x1   0 1 0 0 0 0  0  x1   0   x1 
         
 x2   0 0 1 0 0 0  0  x2   0   x2 
 x3   0 0 0 1 0 0  0  x3   0   x3 
     +  u , y = (1 0 0  0 )  
            
 x   0 0 0 0 0 0  1  xn −1   0  x 
 n −1        n −1 
 xn   − a0 − a1 − a2 − a3 − a4 − a5  − an −1 
 n   b0 
x  xn 

-9-
 Example 3.4: Converting a transfer function with constant term in numerator
Find the state-space representation in phase-variable form equations.
x Ax + Bu
=
C (s) 24 y Cx + Du
=
= 3 → ( s3 + 9s 2 =
+ 26 s + 24)C ( s ) 24 R( s )
R( s ) s + 9 s 2 + 26 s + 24

• D.E. by taking the inverse Laplace transform with zero initial conditions:
D.E.:
c + 9c + 26c + 24c = 24r differential equation

• Select the state variables: x1 = c, x2 = c, x3 = c

⇒ x1 = c, x2 = c, x3 = c

 x1 = x2
 x = x  x1   0 1 0  x1   0 
 2       

3
⇒  x2  =  0 0 1  x2  +  0  r
 x3 = −24 x1 − 26 x2 − 9 x3 + 24r  x   − 24 − 26 − 9  x   24 
 3   3   
 y = c = x
 x1 
 
y = (1 0 0 ) x2 
state-space representation  
 x3 
-10-
Example 3.4: (Continued)

 x   0 1 0   x1   0   x1 
 1       
 x2  =  0
 0 1   x2  +  0  r , y = (1 0 0 ) x2  x Ax + Bu
=
 x   − 24 − 26 − 9   x   24  x 
 3   3     3 y Cx + Du
=

state-space representation

-11-
• Appendix B: Matlab Codes & Outputs '(ch3p3)'
A=[0 1 0;
0 0 1;
-24 -26 -9];
B=[0 0 24]’;
'(ch3p1)’ %B=[0;0;24];
% (ch3p2)
A=[0 1 0; 0 0 1; -9 -8 -7] C=[1 0 0];
C=[2 3 4]
B=[7 8 9]' D=0;
% or
% B=[7; 8; 9]
A1=[0 1 0 F=ss(A, B, C, D)
0 0 1
-9 -8 -7] F =

C = A =
A = x1 x2 x3
2 3 4 x1 0 1 0
0 1 0 x2 0 0 1
0 0 1 x3 -24 -26 -9
-9 -8 -7 B =
B =
7 u1
A1 = 8 x1 0
9 x2 0
0 1 0 x3 24
0 0 1
-9 -8 -7 C =
x1 x2 x3
y1 1 0 0
− state-space model
D =
 x1   0 1 0  x1   0 
       u1
= x2   0 0 1  x2  +  0  r
 x   −24 −26 −9  x   24  y1 0
 3   3   
 x1  Continuous-time state-space model.
 
y = (1 0 0 )  x2 
x 
 3
-12-
 • Appendix B: Matlab Codes & Outputs

'Example 3.4'
num=24;
'Example 3.4’ den=[1 9 26 24];
[A, B, C, D]=tf2ss(num, den);
num=24;
den=[1 9 26 24]; % To 'Phase-variable form'
C (s) 24 [A, B, C, D]=tf2ss(num, den) P=[0 0 1; 0 1 0; 1 0 0];
= 3 2
R( s ) s + 9 s + 26 s + 24 Ap=inv(P)*A*P
Bp=inv(P)*B
• State-Space Representation A = Cp=C*P
-9 -26 -24 Dp=D
1 0 0
x Ax + Bu
= 0 1 0 Ap =
y Cx + Du
= 0 1 0
B = 0 0 1
1 -24 -26 -9
 x1   0 1 0   x1   0  0
       0
 x2  =  0 0 1   x2  +  0  r , Bp =
 x   − 24 − 26 − 9   x   24  0
 3   3    C = 0
0 0 24 1
 x1 
 
=y (1 0 0 )  x2  + 0 ⋅ r D = Cp =
x  0 24 0 0
 3

⇓ Dp =
Controller Canonical Form 0

⇓
Phase-variable Form

-13-
• Decomposing a transfer function

b2
If a3 = 1, +
b1 y
+ +
b0
+ x3 x3 x2 x1
r (t ) 1 ∫ ∫ ∫
− x1 x1 x1
x2
a2

a1
a0

x1 = x2
x2 = x3  x1   0 1 0   x1  0   x1 
⇒  x   0 1   x2 =+ 0  r , y b2 ]  x2 
x3 =
− a0 x1 − a1 x2 − a2 x3 + r
=  2  0 [b0 b1
 x3   −a0 −a1 −a2   x3  1   x3 
y = b0 x1 + b1 x2 + b2 x3
-14-
 Example 3.5: Converting a transfer function with polynomial in numerator
Find the state-space representation of the transfer function.

From Example 3.4:

R( s) s2 + 7s + 2 C (s)
 x1   0 1 0   x1  0 
s 3 + 9 s 2 + 26 s + 24  x   0   x  + 0  r
= 2  0 1  2  
 x3   −24 −26 −9   x3  1 

C ( s ) = (b2 s 2 + b1s + b0 ) X 1 ( s ) = ( s 2 + 7 s + 2) X 1 ( s )

c(t ) =
x1 (t ) + 7 x1 (t ) + 2 x1 (t ) y = c(t ) = b2 x3 + b1 x2 + b0 x1
 x1 = x1 =x3 + 7 x2 + 2 x1

 x1 = x2
state-space representation  =  x1   x1 
x
 2 x
y [b= b1 b2 ]  x2  [ 2 7 1 ]  x2 
3
= 0
 x1   0 1 0   x1  0 
 x 
 2
 0
 0 1   x2  + 0  r  x3   x3 
 x3   −24 −26 −9   x3  1 
 x1 
y = [ 2 7 1 ]  x2 
 x3 
-15-
 Example 3.5: (Continued)

(b)

Figure 3.12
a. Transfer function;
b. decomposed transfer function;
c. equivalent block diagram.
Note: y(t) = c(t)

-16-
 C (s) s2 + 7s + 2
* Controller Canonical Form (page 219) G
= (s) = 3
R ( s ) s + 9 s 2 + 26 s + 24

• Phase-variable Form:
 x1   0 1 0   x1  0   x1 
 x   0   x  + =  r, y
= 2  0 1  2   0 [ 2 7 1]  x2 
 x3   −24 −26 −9   x3  1   x3 

• Renumbering the phase variables in reverse order: x1 → x3 , x2 → x2 , x3 → x1

 x3   x3   0 1 0   x3  0   x3 
 x   0   x  + =  r, y
x =  x2  =  2  0 1    
2 0 [ 2 7 1]  x2 
 x1 
 x1   −24 −26 −9   x1  1   x1 

• Ascending numerical order yields the controller canonical form:

 x1   −9 −26 −24   x1  1   x1 
 x  1   x  + =  r, y 1 7 2  x 
=  2  0 0    
2 0 [ ] 2 
 x3   0 1 0   x3  0   x3 
-17-
* Matrix Transformation for Controller Canonical Form:
ch3p4 (Example 3.4) and procedure (page 219)

 x p AP xP + BP r 
=  xC AC xC + BC r 
=
   
 y = C P xP   y = CC xC 

 xC AC xC + BC r
=

 y = CC xC
0 0 1
  xC= P ⋅ x p
xC =
 0 1 0  xP =
P ⋅ xp →
1 0 0 xC= P ⋅ x p
 

AC PxP + BC r ,
Px p = y=
CC PxP
x p =P −1 AC PxP + P −1 BC r , y =CC PxP
AP BP CP

AP = P −1 AC P
 x p AP xP + BP r
=
 BP = P −1 BC
 y = CP xP CP = CC P
-18-
• Appendix B: Matlab Code for ch3p4 (Example 3.4)

C (s) 24
= 3 2 'Example 3.4' >> T=tf(num, den)
R( s ) s + 9 s + 26 s + 24
num=24;
 numerator  den=[1 9 26 24]; T =
=  24
 denominator  -----------------------
% Controller Canonical Form s^3 + 9 s^2 + 26 s + 24
[A, B, C, D]=tf2ss(num, den);
 x1   0 1 0  x1   0 
Continuous-time transfer function.
      
 x2   0 0 1  x2  +  0  r % To 'Phase-variable form'
 x 
 3
 −24 −26 −9  x   24 
  3    P=[0 0 1; 0 1 0; 1 0 0];
Ap=inv(P)*A*P
 x1  Bp=inv(P)*B AP = P −1 AC P >> [N1, D1]=ss2tf(A,B,C,D)
 
y = (1 0 0 )  x2  Cp=C*P BP = P −1 BC
x  Dp=D N1 =
 3 CP = CC P
0 0 0 24
DP = D
D1 =
Controller Canonical Form Phase-variable Form
1.0000 9.0000 26.0000 24.0000
A = Ap =
-9 -26 -24 0 1 0
1 0 0 0 0 1
0 1 0 -24 -26 -9 >> [N2, D2]=ss2tf(Ap,Bp,Cp,Dp)

B = Bp = N2 =
1 0
0 0 0 0 0 24
0 1
D2 =
C = Cp =
0 0 24 24 0 0 1.0000 9.0000 26.0000 24.0000

D = Dp =
0 0

-19-
3.6 Converting from State Space to a Transfer function

• State and output equations: x = Ax + Bu, y = Cx + Du

• Laplace transform with zero sX( s ) = AX( s ) + BU( s )

initial conditions: Y( s ) = CX ( s ) + DU( s )

• Solving for X(s): ( sI − A)X( s ) = BU( s )

X( s ) = ( sI − A) −1 BU( s )

• Solving for Y(s): Y( s) = C ( sI − A) −1 BU( s ) + DU( s )

[ ]
= C ( sI − A) −1 B + D U( s )

• If U(s) and Y(s) are scalars, Y ( s)

T ( s) = = C ( sI − A) −1 B + D
the transfer functions is: U ( s)

-20-
Example 3.6: State-space representation to transfer function
Find the transfer function, T(s)=Y(s)/U(s).

0 1 0  10 
    Y ( s)
x =  0 0 1  x +  0 u , y = (1 0 0 ) x ⇒ T ( s) = = C ( sI − A) −1 B + D ?
  0 U ( s)
 −1 − 2 − 3  

 s 0 0  0 1 0   s −1 0 
     
( sI − A) =  0 s 0  −  0 0 1  = 0 s −1 
 0 0 s   −1 − 2 − 3  1 2 s + 3
     

 s 2 + 3s + 2 s+3 1
 
 −1 s ( s + 3) s
 2
adj( sI − A ) − s − ( 2 s + 1) s 
( sI − A) −1 = = 3 2
det( sI − A) s + 3s + 2 s + 1

Y (s)
T (s) = = C ( sI − A) −1 B + D
U (s)
10 
−1   10( s 2 + 3s + 2)
= (1 0 0 )(sI − A )  0  + 0 = 3 ⇐ The final result!
0 s + 3s 2 + 2 s + 1
 
-21-
• Using Matlab to convert a state-space representation to a transfer function

num =
0 10.0000 30.0000 20.0000
den =
A=[0 1 0;0 0 1;-1 -2 -3]; 1.0000 3.0000 2.0000 1.0000
B=[10; 0; 0];
C=[1 0 0];
D=0; a =
x1 x2 x3
x1 0 1 0
[num,den]=ss2tf(A,B,C,D,1) x2 0 0 1
%G(s)=num/den, x3 -1 -2 -3

b =
Tss=ss(A,B,C,D) u1
x1 10
x2 0
'Polynomial form, Ttf(s)' ans = Polynomial form, Ttf(s)
x3 0
Ttf=tf(Tss) Transfer function: c =
10 s^2 + 30 s + 20 x1 x2 x3
'Factored form, Tzpk(s)' --------------------- y1 1 0 0
Tzpk=zpk(Tss) s^3 + 3 s^2 + 2 s + 1

ans = Factored form, Tzpk(s) d =

u1
y1 0
Zero/pole/gain:

10 (s+2) (s+1)
Continuous-time model.
-----------------------------------
(s+2.325) (s^2 + 0.6753s + 0.4302)

-22-
• Using Matlab’s Symbolic Math Toolbox to convert a state-space
representation to a transfer function

'(ch3sp1) Example 3.6' % Display label.

syms s % Construct symbolic object for frequency
% variable 's'.
A=[0 1 0;0 0 1;-1 -2 -3]; % Create matrix A.
B=[10;0;0]; % Create vector B.
C=[1 0 0]; % Create vector C.
D=0; % Create D.
I=[1 0 0;0 1 0;0 0 1]; % Create identity matrix.
'T(s)' % Display label.
T=C*((s*I-A)^-1)*B+D; % Find transfer function.
pretty(T) % Pretty print transfer function.

ans =
(ch3sp1) Example 3.6

ans =

T(s)
2
s + 3 s + 2
10 -------------------
3 2
s + 3 s + 2 s + 1
>>

-23-
 Walking robots (Hannibal): explore hostile
3.7 Linearization environments and rough terrain

Example 3.7: Representing a nonlinear system

Linearize the state equations about the pendulum’s
equilibrium point

© Bruce Frisch/S.S./Photo Researchers

Torque angular displacement

Spring: T ( s ) = Kθ ( s )
Viscous damper: T ( s ) = Dsθ ( s )
Inertia: T ( s ) = Js 2θ ( s )

d 2θ MgL
J 2 + sin θ = T  x1 = x2
dt 2 
⇒  MgL T
dθ x
 2 = − sin x +
Let x1 = θ , x2 = 2J
1
J
dt
d 2θ MgL T
2
=
− sin θ +
dt 2J J
-24-
• We need to linearize the equation about equilibrium point. dθ
x1 = θ , x2 =
dt
dθ
• The equilibrium point is: x1 = 0, x2 = 0 → θ = 0, =0
dt
x  θ 
⇒  1  =    = 0
 x2  e  θ  e

• Let x1 and x2 be perturbed about the equilibrium point:

x1= 0 + δ x1   x1   0   δ x1 
 ⇒  =  + 
x2= 0 + δ x2   2   0   δ x2 
x

• Taylor series expansion of sin(x) about equilibrium point:

df ( x1 ) ( x1 − x0 ) d sin( x1 )
f ( x=
1) sin( x1 ) ≈ f ( x0 ) + +  ≈ sin(0) + 0) δ x1
( x1 −=
dx x = x0 1! dx1 x1 = 0

 x1 = x2  δ x1 = δ x2  0 1 0

   δ x1     δ x   
 MgL T ⇒  MgL = T 1
+T 
    − MgL 
 δ x2 = (δ x1 ) +
x = − sin x +  −  δ x2  
0   δ x2   
 2 2J
1
J 2J J  2J  J

-25-