EEM 342: Fundamentals of Control Systems

2006-2007 Spring Semester 1st Midterm Exam

Time: 120 minutes; Total Points: 100
One size-A4 formula sheet and a hand calculator, not capable of symbolic operations, may be used.

0. (1 point) Write the group you are registered to on the top-right corner of your answer sheet.

1. Consider the system described by the following state equations:

" # " #
−1 1 1
ẋ(t) = x(t) + u(t)
0 −2 0
" # " # " #
y1 (t) 1 1 0
y(t) = = x(t) + u(t)
y2 (t) 1 0 1

a) (9 points) Is the system (i) linear, (ii) time-invariant, (iii) causal? Explain your answer in
each case.
" #
1
b) (10 points) Let x(1) = and u(t) = e−t for t ≥ 1. Find x(t) and y(t) for t ≥ 1.
−1
c) (10 points) Find the transfer function matrix of the system.
d) (10 points) Find the (two-variable) impulse response matrix of the system.
e) (10 points) Let the(system be initially relaxed. Find the output y(t), for −∞ < t < ∞, in
1 , |t| ≤ 1
response to u(t) = .
0 , |t| > 1
f) (10 points) Determine the minimum time constant, τ min , of the system. Then, find the
state-space equations of the discrete-time equivalent of the system, where the sampling
period (or integration step size) is τmin
10 .

2. (20 points) Write the state equations of a minimal realization of the transfer function

s3 + 3s2 + 3s + 1
G(s) = .
s3 + 3s2 + 2s
3. (20 points) Consider the system shown below, where S 1 is the system given in Question 1 and
S2 is the system given in Question 2. Find the transfer function from r to v.

y1 - v
+ u
r - - S1

6− y2

S2 
 SOLUTIONS

1. a) (i) The system is linear since the state equations (both the dynamics equation and the
output equation) are linear in x and u.
(ii) The system is time-invariant since the state equations (more precisely the coefficient
matrices) do not depend on the time variable t.
(iii) The given equations can be solved either forward in time or backward in time. If they
are solved forward in time (which is the presumed case) then x and y do not depend
on future values (they depend on past and present values) of x and u, hence the system
is causal. If the equations are solved backward in time then x and y in general depend
on present and future values (they do not depend on past values) of x and u, hence, in
this case, the system is non-causal (in fact, it is anti-causal).

" # " # " # " #

−1 1 1 1 1 0
b) With A = ,B= ,C= ,D= , we have
0 −2 0 1 0 1
" #−1  " #
s + 1 −1 e−t e−t − e−2t
e At
= L−1  =
0 s+2 0 e−2t

Then
" #
t e−2(t−1) + (t − 1)e−t
Z
x(t) = eA(t−1) x(1) + eA(t−τ ) Bu(τ )dτ = , t≥1
1 −e−2(t−1)

and " #
(t − 1)e−t
y(t) = Cx(t) + Du(t) = , t≥1
e−2(t−1) + te−t

 
1
s+1
c) G(s) = C(sI − A)−1 B + D =  .
s+2
s+1


 0 , t<0


−1
" #
d) Go (t) = L [G(s)] = e−t
 , t≥0
e−t + δ(t)



Thus, 
 0 , t<τ


" #
G(t, τ ) = Go (t − τ ) = e−(t−τ )
 , t≥τ
e−(t−τ ) + δ(t − τ )



The same answer can also be obtained directly from

(
0 , t<τ
G(t, τ ) = A(t−τ )
Ce B + Dδ(t − τ ) , t ≥ τ

1
 e) For t < −1, u(t) = 0. Therefore, since the system is initially relaxed, by causality, y(t) = 0,
t < −1.
For −1 ≤ t ≤ 1, u(t) = 1. Thus,
" # " #
∞ t e−(t−τ ) 1 − e−t−1
Z Z
y(t) = G(t, τ )u(τ )dτ = dτ = , −1 ≤ t ≤ 1
−∞ −1 e−(t−τ ) + δ(t − τ ) 2 − e−t−1

For t > 1, u(t) = 0. Thus,

" # " #
∞ 1 e−(t−τ ) (e1 − e−1 )e−t
Z Z
y(t) = G(t, τ )u(τ )dτ = dτ = , t>1
−∞ −1 e−(t−τ ) + δ(t − τ ) (e1 − e−1 )e−t

Therefore,  " #
 0
, t < −1



0




 " #
1 − e−t−1


y(t) = , −1 ≤ t ≤ 1


 2 − e−t−1
 " #


 (e 1 − e−1 )e−t


 1 −1 −t
, t>1

(e − e )e

1
f) The eigenvalues of the dynamics matrix are σ(A) = {−1, −2}. Thus, τ min = max{1,2} = 12 .
1
Therefore, we take the the sampling period as τ s = 20 = 0.05. Then
" # " #
Aτs e−τs e−τs − e−2τs 0.9512 0.0464
Ad = e = =
0 e−2τs 0 0.9048
" # " # " #
τs τs e−τ 1 − e−τs 0.0488
Z Z
Aτ
Bd = e Bdτ = dτ = =
0 0 0 0 0

Therefore, the discrete-time equivalent system is described as:

" # " #
0.9512 0.0464 0.0488
xk+1 = xk + uk
0 0.9048 0
" # " #
1 1 0
yk = xk + uk
1 0 1
where xk := x(kτs ), uk := u(kτs ), and yk := y(kτs ).

s3 + 3s2 + 3s + 1 (s + 1)3 (s + 1)2 1

2. G(s) = 3 2
= = = 1+ 2 . By defining v̂(s) =
s + 3s + 2s" s(s
#
+ 1)(s + 2) s(s + 2) s + 2s
1 v(t)
2
û(s) and x(t) = , the following state equations are obtained:
s + 2s v̇(t)
" # " #
0 1 0
ẋ(t) = x(t) + u(t)
0 −2 1
h i
y(t) = 1 0 x(t) + u(t)

2
3. The transfer function matrix of S 1 was found in Question 1 part c. The system has the following
block diagram:

- 1 -
s+1 v
+
r -

6−
- s+2
s+1

(s + 1)2 
s(s + 2)

This diagram can be reduced to:

+ 1
r - - 1 - - v
 s+1
6−

s+1 
s

which can further be reduced to:

s 1
r - - - v
2s + 1 s+1

Therefore, the transfer function matrix from r to v is:

1 s s
  
G(s) = =
s+1 2s + 1 2s2 + 3s + 1