Tenta13a 2

Exam Maj 23 2013 in SF2852 Optimal Control.
Examiner: Johan Karlsson, tel. 790 84 40.

Allowed books: The formula sheet and β mathematics handbook.
Solution methods: All conclusions should be properly motivated.
Note! Your personal number must be stated on the cover sheet. Number
your pages and write your name on each sheet that you turn in!
Preliminary grades (Credit = exam credit + bonus from homeworks):
23-24 credits give grade Fx (contact examiner asap for further info), 25-27
credits give grade E, 28-32 credits give grade D, 33-38 credits give grade C,
39-44 credits give grade B, and 45 or more credits give grade A.
1. Solve the following optimal control problem using PMP

Z 1 (
ẋ = −x + u
min u2 (t)dt subject to
0 x(0) = 1, x(1) = 0
The optimal control as well as the optimal cost should be given. (10p)
2. Consider the scalar linear quadratic optimal control problem

Z ∞
min (x2 + u2 )dt subject to ẋ = −x + u, x(0) = 1 (1)
0
(a) Compute the optimal stabilizing feedback control and the corre-
sponding optimal cost. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (3p)
(b) Compute the closed loop poles. . . . . . . . . . . . . . . . . . . . . . . . . . . . (1p)
Now consider the finite truncation of (1)

Z T
min (x2 + u2 )dt subject to ẋ = −x + u, x(0) = 1 (2)
0
(c) Use the Hamilton-Jacobi-Bellman equation to compute the opti-

mal feedback control and the corresponding optimal cost. . (4p)
(d) Let p(t, T ) be the Riccati solution corresponding to (2), where
the final time is made explicit as an argument. Compute
lim p(t, T )
T →∞
and compare with the solution to the ARE corresponding to (1).

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2p)
1
3. The path of a moving target is estimated using a radar. At a number
of time points t ∈ {1, . . . , T } measurements are taken. Each measure-
ment results in a set of nt positions Mt = {xt1 , xt2 , . . . , xtnt }. One of
the positions in Mt is the correct position of the target at that time,
whereas the other time points correspond to scatters from other ob-
jects. The goal is to determine which set of positions that give the
most likely positions of the object.
The target is moving according to a Gaussian distribution. Given that
the position of the target is xt at time t, the probability distribution
of the target in time t + 1 is given by

1 1 2
p(xt+1 |xt ) = √ exp − (xt − xt+1 ) .
2π 2
The likelihood of a path x1 , x2 , . . . , xT is then given by

−1
TY
L(x1 , . . . , xT ) = p(xt+1 |xt ).
t=1
The problem is to find the path (x1 , . . . , xT ) that has the highest like-
lihood out of the paths that are consistent with the measurements,
i.e., xt ∈ {xt1 , xt2 , . . . , xtnt } for all t = 1, . . . , T .
(a) Formulate the problem as a dynamic programming problem. (Hint:

formulate it in terms of the log-likelihood.) . . . . . . . . . . . . . . . . (6p)
(b) Use dynamic programming to find the most likely path given the
following set of measurements (with T=4).
M1 = {2, 4},
M2 = {2, 5, 8},
M3 = {4, 6, 8},
M4 = {2, 7, 8}.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4p)
4. Consider the optimal control problem

Z ∞
min (u − x)2 dt subject to ẋ = ax + u, x(0) = x0 .
0
(a) Find the optimal stabilizing solution (i.e., x(t) → 0 as t → ∞)

for the case a > −1. Determine the optimal cost. . . . . . . . . . . (6p)
(b) What is the optimal control when a < −1. . . . . . . . . . . . . . . . . . (3p)
(c) Discuss the optimal control when a = −1. . . . . . . . . . . . . . . . . . . (1p)
2
5. Consider the following optimal control problem



 ẋ = −x2 + u
Z T 
x(0) = 0
max x(t)dt subject to
0 

 0 ≤ u(t) ≤ 1
R T u(t)dt = K,

0
where T > K > 0 are given.

R
(a) Introduce the extra state y = udt. Define the Hamiltonian and
the TPBVP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2p)
(b) Determine the possible optimal controls u. . . . . . . . . . . . . . . . . (3p)
(c) Draw the phase diagram in (x, λ) and study it to determine the
possible switching sequences. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .(3p)
(d) Find T > K > 0 such that the control
(
1 t<K
u(t) =
0 t≥K
is not optimal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2p)

Tenta13a 2

Uploaded by

Copyright:

Available Formats

Tenta13a 2

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tenta13a 2

Uploaded by

Copyright:

Available Formats

Exam Maj 23 2013 in SF2852 Optimal Control.

Examiner: Johan Karlsson, tel. 790 84 40.

1. Solve the following optimal control problem using PMP

2. Consider the scalar linear quadratic optimal control problem

Now consider the finite truncation of (1)

(c) Use the Hamilton-Jacobi-Bellman equation to compute the opti-

and compare with the solution to the ARE corresponding to (1).

The likelihood of a path x1 , x2 , . . . , xT is then given by

(a) Formulate the problem as a dynamic programming problem. (Hint:

4. Consider the optimal control problem

(a) Find the optimal stabilizing solution (i.e., x(t) → 0 as t → ∞)

where T > K > 0 are given.

is not optimal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2p)

You might also like