Final Exam Automatic Control II Reglerteknik II 5hp: Good Luck!
Final Exam Automatic Control II Reglerteknik II 5hp: Good Luck!
Final Exam Automatic Control II Reglerteknik II 5hp: Good Luck!
Automatic Control II
Reglerteknik II 5hp
Preliminary grades: 23p for grade 3, 33p for grade 4, 43p for grade 5.
Use separate sheets for each problem, i.e. no more than one problem per
sheet. Write your exam code on every sheet.
Good luck!
Problem 1 Please use English in your solution for Problem 1.
(a) Give a state space representation for the continuous-time system
h i
Y (s) = 1 2s+3 U 1 ( s) :
s+3 (s+3)(s+1) U2 (s)
(4p)
(b) The continuous-time system
0 1
ẋ(t) = 1 0
x(t) + 10 u(t);
y (t) = 1 0 x(t);
is sampled (zero-order-hold) with sampling period h, and the discrete-time
model
(
x(kh + h) = F x(kh) + Gu(kh); cos h sin h
with F =
y (kh) = Hx(kh); sin h cos h
is minimized.
(a) Show that the system is unstable. (1p)
(b) Find the feedback gain L in the optimal control law u(t) = Lx(t). (4p)
(c) What will the pole of the closed loop system be as Q ! 0? (3p)
1
Problem 3 Consider the (scalar) continuous-time system
˙ (t) = u(t) + w(t);
y (t) = (t) + e(t);
where e(t) is white noise with intensity Φe (! ) = 0:1, and w (t) is a disturbance
described by
2 1
˙(t) = 1 0
(t) + 10 (t);
w(t) = 1 0 (t);
where (t) is white noise with intensity Φ (! ) = 1.
(a) Determine the spectral density for w (t), i.e. Φw (! ). In particular, what
is Φw (0), Φw (1) and lim Φw (! )? (2p)
!!1
(b) Determine the covariance matrix Π = E (t) T (t) for the state vector in
the disturbance model. (3p)
(c) The system model and the disturbance model can be combined into an
augmented state space model (the “standard” form)
ẋ(t) = Ax(t) + Bu(t) + Nv1 (t);
y (t) = Cx(t) + v2 (t);
where v1 (t) and v2 (t) are white noise with intensities Φv1 (! ) = R1 and
Φv2 (! ) = R2 . Find such an augmented state space model, that is define
the state variables in x(t), determine the matrix A and the vectors B , N and
C accordingly, and give the values of the noise intensities R1 and R2 . (4p)
2
Problem 5 In a water tower drinking water is stored at a height sufficient to
pressurize the water distribution system. The water level should be constant.
The water supply is monitored by measurements of the water level. This
system can be modeled as (the discrete-time system)
x(k + 1) = x(k) + v1 (k);
y (k) = x(k) + v2 (k);
where v1 (k ) describes the fluctuations in the net flow into the water container,
and v2 (k ) is the measurement noise. Both v1 and v2 are white noise, and have
the intensities R1 = 1, R2 = 0:5 and R12 = 0:5. In order to estimate the
water level a Kalman filter is used:
x̂(k + 1jk) = x̂(kjk 1) + K (y (k) x̂(kjk 1)):
(a) Determine the covariance of the estimation error P = E x̃2 (k jk 1). (3p)
(b) What is the Kalman gain K ? (2p)
(c) Find the transfer function from the measured output y (k ) to the esti-
mated level x̂(k jk 1). What is the static gain? (2p)
(d) A Kalman filter is designed for a third order system. One of the following
matrices,
2 3
1 0 0
1 0 1 0 0
M1 =
0 2
; M2 4
= 0 2 15 ; M3 =
0 2 1
;
0 1 2
2 3 2 3
1 0 0 1 0 0
M4 4
= 0 2 15 ; M5 4
= 0 2 15 ;
0 1 2 0 1 2
is the covariance matrix P of the estimation error x̃. Which of the matrices
is P ? A full motivation is required! (3p)
3
Solutions to the exam in Automatic Control II, 2011-10-24:
0 = 2S + Q S2 , S2 2S Q = 0;
p
with solution S = + p 2 + Q (the negative squareroot is rejected since
S 0). Thus L = S = + 2 + Q.
(c) The closed loop system is
p
ẋ = x Lx = ( L)x = 2 + Qx;
1
p
with the pole 2 + Q ! as Q ! 0. (This means that the “cheapest”
— only u is penalized — way to stabilize the system is to mirror the unstable
pole.)
3. (a) We have that Φw (! ) = jGw (i!)j2Φ (!) if w(t) = Gw (p) (t). Here
Φ (! ) = R = 1, and
p p
Gw (p) = = ;
p + 2p + 1 (p + 1)2
2
2
5. (a) The covariance P = PT 0 is the solution of the DARE
P = F P F T + NR1 N T (F P H T + NR12 )(HP H T + R2 ) 1 (F P H T + NR12 )T :
Here F = 1, N = 1, H = 1, R1 = 1, R2 = 0:5 and R12 = 0:5 )
(P + 0:5)2
P =P +1+
P + 0:5
, (P + 0:5)2 = P + 0:5 , P = 0:5;
where the negative solution is rejected. Thus P = 0:5.
(b) K = (F P H T + NR12 )(HP H T + R2 ) 1 ) K = 00::5+0 :5
5+0:5
= 1.
(c) We have
6. (a) In order to sample the system, represent it in state space form, e.g.
in controller canonical form:
ẋ = 2 0
x + 10 u; ) F =e Ah
=
e 2h
0
=
e 1
0
;
1 0 0:5(1 e 2h
) 1 0:5(1 e 1
) 1
y = 0 1 x; H=C= 0 1 and
Z h
G= eAt
Bdt = 0:5(h0:5(1 e 2h )
0:5) + 0:25e 2h =
0:5(1 e 1 )
0:25e 1
:
0
The transfer operator is then
G (q ) = H (qI F ) G = 0 1
d 1 q e 1
0
1
0:5(1 e 1 )
0:5(1 e ) q 1
1
0:25e 1
0:25(1 e 1 )2 + 0:25e 1 (q e 1 ) 0:25e 1 (q + e 2)
= = :
(q 1)(q e 1 ) (q 1)(q e 1 )
(b) The poles of the closed loop system are given by