You are on page 1of 5

Control Systems Theory and Design

Exam date: 22.07.2008

Problem 1 (25 Marks)

f1
z
f4
θ ω1
f3
φ x f2

ω3
ω4
y
ψ
ω2

Figure 1: Quadrocopter

A quadrocopter as shown in Figure 1 has four propellers rotating with velocities ω1 , ω2


counterclockwise and ω3 , ω4 clockwise, respectively. The forces f1 , . . . , f4 applied to the
frame are considered as control inputs. The measured outputs are the angles φ, θ and ψ,
which describe the orientation, and the altitude z. The simplified equations of motion are

Jφ φ̈ = l (f2 − f1 ),
Jθ θ̈ = l (f3 − f4 ),
Jψ ψ̈ = c (f1 + f2 − f3 − f4 ),
m z̈ = f1 + f2 + f3 + f4 ,

with constants l, c, overall mass m, and the moments of inertia Jφ , Jθ and Jψ . In the
following assume that l = 1, c = 1, m = 1, Jφ = 1, Jθ = 1 and Jψ = 1.

a) Define a state vector for the quadrocopter. Hint: Arrange the states such that a
variable is directly followed by its derivative (if needed).
Derive a state space model of the quadrocopter corresponding to your choice of state
vector.

b) What are the eigenvalues of the state space model derived in (a)?
What is the meaning of these eigenvalues in terms of the physical model of the
quadrocopter?

c) Are the states of the quadrocopter observable from the measured outputs?

d) The quadrocopter state-space matrices A, B, C and D are in the Matlab workspace.


Give Matlab code for designing a state-feedback LQR controller and for plotting
the closed loop step responses.

1
Problem 2 (25 Marks)

A plant is modeled by the transfer function matrix


 
3 5
 s+4 s2 +3s+2 
G(s) =  .
s+1 1
s2 +5s+4
− s+2
A state space realization of the same plant is given by

ẋ (t) = Ax (t) + Bu (t) ,


y (t) = Cx (t) + Du (t) .

where
   
−4 0 0 0 0 0 2 0
 0 −5 −2 0 0 0 1 0
   
0 2 0 0 0 0 0 0

A= , B= ,
0 0 0 −3 −2 0  0 2
   
0 0 0 1 0 0 0 0
0 0 0 0 0 −2 0 1

· 3 5
¸ · ¸
0 0 0 0 0 0
C= 2
1
2 , D= .
0 1 2
0 0 −1 0 0

a) Has the above state space realization of G(s) a minimum number of state variables?
If not, what is the minimal number of state variables of this system?
© ª
b) Is any of the numbers −3 ; − 83 ; 0 a transmission zero of G(s)? State reasons for
your answer.

c) By analyzing the above state space model with the Matlab function zero it turns
out that {−2} is also a zero of the state space model of the plant. Will this zero
have an effect on the input-output behaviour of the system?

d) Give Matlab code to determine the Kalman decomposition of the system above.
Assume that the system matrices (A, B, C and D) are available in the workspace.

2
Problem 3 (25 Marks)

Let a continuous-time system G(s) be given as :


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

a) Find the pulse transfer function H(z) corresponding to G(s) using the Tustin ap-
proximation with sampling time T = 1 sec.

b) How many zeros does H(z) have? Among these how many are sampling zeros?

c) Compute the step response of H(z) at t = 0 and t = 1.

d) Let G(s) be in the Matlab workspace as G. Give Matlab code to find the exact
discretization of G(s) for the sampling time T = 1 sec.

e) Consider the poles or pole pair locations a, b, c, d and e in the z-plane marked in
Figure 2 (see next page). Find the corresponding impulse responses in Figure 3.

3
b.
1

0.8

0.6

0.4

0.2
imaginary

c. d.
a. e.
0

−0.2

−0.4

−0.6

−0.8

−1

−1 −0.5 0 0.5 1 1.5 2


real

Figure 2: Pole locations in z-plane.

i. 20 iv. 1

10 0

0 −1
0 1 2 3 4 5 0 1 2 3 4 5

ii. 1 v. 1

0 0

−1 −1
0 1 2 3 4 5 0 1 2 3 4 5
iii. 1

−1
0 1 2 3 4 5

Figure 3: Impulse responses of discrete-time systems.

4
Problem 4 (25 Marks)

The impulse response of an unknown system is measured, and a state-space model of the
system is to be identified in the form

x(k + 1) = Φ x(k) + Γ u(k)


y(k) = C x(k) + D u(k)

a) Give the values of the impulse response of the system for k = 0, 1, 2, 3 in terms of
the state space model.

b) Explain in detail how one can determine the order of the system from an ideal
(noise- and error-free) measurement of the impulse response of the system?

c) From a noisy measurement of the impulse response, the Hankel matrix H is con-
structed, and its singular values ordered by magnitude are

σi = {54.31, 23.72, 22.10, 0.56, 0.09, 0.01, ...}

c.1) What system order would you estimate? Reasons!


c.2) Write Matlab code to compute Γ and C, when the Hankel matrix H is in the
workspace. The system has 2 inputs and 2 outputs.

d) Assume now that the extended controllability and observability matrices are in the
workspace as Ctr and Obs, respectively. Which of the following Matlab scripts can
be used to compute the system matrix Φ (in Matlab Phi)? Give reasons.

A) obs0 = Obs(1:2, :);


obs1 = Obs(3:4, :);
Phi = inv(obs1) * obs0

B) ctr0 = Ctr(:, 1:end-2);


ctr1 = Ctr(:, 3:end);
Phi = ctr1*ctr0’ * inv(ctr0*ctr0’)

C) obs0 = Obs(1:end-2, :);


obs1 = Obs(3:end, :);
Phi = obs0\obs1

D) obs0 = Obs(1:end-2, :);


obs1 = Obs(3:end, :);
Phi = inv(obs1) * obs0

You might also like