Exam RT2 2015 Summer
Exam RT2 2015 Summer
Exam RT2 2015 Summer
Rq L1 C1
iq iL
vR v1 v2 i1 i2
vq C2 v3 L2 v4 vL RL
ẋ = A x + b u,
(1)
y = c x.
T
with the state vector x = iq i2 vR v2 vL of the system depicted in Figure 1.
Assume that vq is the input, vL is the output (i.e. u(t) = vq (t), y(t) = vL (t)) and
L1 = L2 = 1 H, RL = 1 Ω,
(2)
C1 = C2 = 1 F, Rq = 3 Ω.
Hints:
1
4. Current-voltage relation:
C L iL
iC dvC diL
iC = C (5) vL = L (6)
dt dt
vC vL
Apply the transformations to b, c and x, and write down the new state space
representation
x̄˙ = Ā x̄ + b̄ u,
(9)
y = c̄ x̄.
2
Problem 2 (25 Marks)
Hint: Tasks (a) and (b), see next page, can be solved independently.
Consider a system described by the block diagram shown in Figure 4.
G1 (s)
u1 e1 y1
G2 (s)
G3 (s)
y2 e2 u2
The transfer functions G1 (s), G2 (s) and G3 (s) of the subsystems are given by:
−s 1
s+3 s+3
1 2
G1 (s) = , G2 (s) = , G3 (s) = s−1 .
s+3
1.5 s
s+1 s+1
u1 → y 2
u2 → y 2
3
b) Figure 5 shows an observer-based state feedback controller augmented with an in-
tegrator block.
r 1 xI u y
s
FI G(s)
−
x̂ Observer
ẋ = Ax + Bu
y = Cx
T
i) Let x̃ = x − x̂ denote the estimation error. Given the state vector x x̃ xI ,
calculate the state matrices Acl , Bcl , Ccl . Write them in terms of A, B, C, F, FI , L.
Hint: use x̃˙ = (A + LC)x̃ and x̂ = x − x̃.
ii) What happens to x̃ when a step input r(t) = σ(t) is applied? (Assume x̂(0) =
x(0) = 0).
iii) Give MATLAB code to calculate the optimal state feedback gain F and integral
gain FI . Give reasonable values (with appropiate dimensions) to the weighting
matrices Q and R. Assume A ∈ R3×3 , B ∈ R3×2 , C ∈ R2×3 , D = 02×2 are
already in the workspace.
4
Problem 3 (25 Marks)
and both observer-based state feedback configurations shown in Figure 6 and Figure 7.
u y
G(s)
x̂ −
Observer r
r 1 2 y
v × × G(s)
x̂
Observer
r 1 xI u y
s
FI G(s)
−
x̂ Observer
5
a) i) Calculate a state feedback gain f that places the poles of A + bf at −3 −5 .
Hint: exploit the fact that the model is in controller canonical form.
ii) From the following list, choose the most appropiate values for the poles of
A + lc.
a) −15 −25 . b) −3 · 107 −5 · 107 . c) −3 −5 .
iii) Calculate an appropiate value for v in Figure 7, so that the static gain r → y
is 1.
b) i) Copy the loop in Figure 7 in your answer sheet and draw a box around the
controller. How many inputs does the controller have?
ii) Figure 9 shows step responses of the loops in Figure 6 and Figure 7. Which
response belongs to which configuration? Give reasons for your answer.
2
B
1.5
A
1
0.5
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
c) i) Compare the loops in Figure 7 and Figure 8. Both are observer-based con-
trol loops with reference tracking capabilities. Name one advantage and one
disadvantage of the loop in Figure 7.
ii) In Figure 7, two break points (labeled 1 and 2) have been marked. Where
would you expect an input disturbance of a real system to enter the loop?
Give reasons for your answer.
Hint: use your block diagram from (b)(i).
iii) Assume a disturbance d1 enters at 1 in Figure 7, and a disturbance d2 enters
at 2. Which of these disturbances has an effect on the estimation error? Give
reasons for your answer.
d) Complete the following MATLAB code to build the closed loop in Figure 7. Assume
that A, b, c and vectors vec f and vec l containing the desired pole locations for
A + bf and A + lc, respectively, are given in the workspace.
1 G= ...
2 f= ...
3 l= ...
4 v= ...
5
6
Problem 4 (25 Marks)
and assume it is being discretized using both exact (zero-order hold) discretization
and Tustin (bilinear) approximation, leading to Gzoh (z) and GTustin , respectively.
Answer the following questions:
i) How many sampling zeros does the transfer function Gzoh (z) have?
ii) How many sampling zeros does the transfer function GTustin (z) have?
iii) What type of zeros would you expect, when G(s) is sampled at high sampling
rates with a zero-order-hold?
iv) Sketch the poles of G(s) in the s-plane for β = −3 and map them into the z-
plane using exact discretization (zero-order-hold) with sampling time T = 2 sec.
c) Give MATLAB commands to calculate Gzoh (z) and GTustin (z) with a sampling time
of T = 2 sec. Assume that α = 2, β = −3. The workspace is empty.