Optimal Jan24 Assignment 3
Optimal Jan24 Assignment 3
1. Consider a train moving on a railway. The problem is to drive the train to a station and stop it there in a
minimal time. Describe position of the train by a coordinate x1 on the real line. Assume that the train moves
without friction, and we can control acceleration of the train by applying a force bounded by absolute value.
Using rescaling if necessary, we can assume that absolute value of acceleration is bounded by 1 i.e. u ∈ [−1, 1]
ẋ1 = x2
ẋ2 = u (1)
Let us consider the problem of getting to the origin (0, 0) (desired station) from the initial state x0 (initial
station) in the shortest time for constrained input u ∈ [−1, 1]. Find out the time-optimal control input. Draw
the phase portrait on application of the derived control input for any arbitrary initial condition x0 . State your
arguments clearly.
2. Consider a linear oscillator whose motion can be controlled by force bounded in absolute value. The consider
corresponding control system
ẋ1 = x2
ẋ2 = −x1 + u (2)
where u ∈ [−1, 1]. We consider the time-optimal problem for this system:
x(0) = x0 , x(t1 ) = 0
t1 −→ min
Find the time-optimal control input. Draw the phase portrait on application of the derived control input for
any arbitrary initial condition x0 . State your arguments clearly.
3. (Dubin’s car problem) Consider a car moving in the plane. The car can move forward with a fixed linear
velocity and simultaneously rotate with a bounded angular velocity. Given initial and terminal position and
orientation of the car in the plane, the problem is to drive the car from the initial configuration to the terminal
one for a minimal time. A simplified model for car is given as follows:
ẋ1 = cos(θ)
ẏ1 = sin(θ)
θ̇ = u (3)
λd (t) = Ki(t) is the instantaneous torque developed; K = kt Ia is the product of the torque constant and the
armature current, and I is the angular moment of inertia. Refer to Figure 1. The performance measure to be
minimised
Z ∞
i2 (t) + ω 2 (t) + vin
2
J= (t) dt (4)
0
and the admissible controls are not bounded. Consider R, L, K, I equal to 1 and find the optimal control.
∗ Asst. Professor, Electrical Engineering, Office: EE 214D, e-mail: dm@ee.iitb.ac.in
1
Figure 1: DC motor
ẋ = −ax + u.
It is desired to bring the system from arbitrary fixed initial state x(0) = x0 to the origin in T seconds and
minimize the performance measure
Z T
J= u2 dt
0
ẋ1 =x2
ẋ2 = − x2 + u
ẋ = −0.1x + u
where x(t) is the height of the water, and u(t) is the net inflow rate of water at time t. Assume 0 ⩽ u(t) ⩽ M .
a. Find the optimal control law if it is desired to minimize
Z 100
J= −x(t) dt
0
ẋ = −x + u
to be transferred to the origin from an arbitrary initial state. The performance measure is
Z 1
1 2
J= [3x + u2 ] dt.
0 2
2
10. (Free time fixed endpoint) Assume the controls are constrained to lie in the interval [0, 1]. We want to minimize
RT RT
J(u) = 0 L(x, u, t)dt = 0 ut dt, where u(t) ∈ [0, 1] for all t ∈ [0, T ] and the system states satisfy x˙1 = 1 − 2u
and x2 = −1 with fixed initial and terminal conditions x1 (0) = 0,x2 (0) = h and x1 (T ) = 0,x2 (T ) = 0.(Terminal
time T is free.)
11. Consider
R 2the2 dynamical system given by x˙1 = x2 and x˙2 = −2x1 − 3x2 + u. The running cost is given by
J(u) = 0 x1 + x2 + 0.5u2 dt. What would be the minimum cost to reach point (2, 1) from initial position (8, 0).
2
12. ConsiderR the dynamical system given by x˙1 = x2 and x˙2 = −2x1 − 3x2 + u. The running cost is given by
2
J(u) = 0 x21 + x22 + 0.5u2 dt. If one required to take system from initial position (8, 0) to the circle given by
R2
(x1 − 0)2 + (x2 − 1)2 = 22 while minimizing J(u) = 0 x21 + x22 + 0.5u2 dt, then what would be the optimal
control that would perform this task.
13. Find the optimal states, Co-states and control input for the given system: ẋ1 = x2 , ẋ2 = u(t) and the functional
2
u2
Z
J= dt
0 2
for the given boundary conditions: x(0) = [1, 2]T and x(2) = [1, 0]T .
14. Find the optimal states, Co-states and control input for the given system: ẋ1 = x2 , ẋ2 = u(t) and the functional
2
u2
Z
J = 0.5 ∗ (x1 (2) − 4)2 + 0.5 ∗ (x2 (2) − 2)2 + dt
0 2
for the given boundary conditions: x1 (0) = 1, x2 (1) = 2 and x1 (2) = 0, x2 (2) = F ree .
15. Show that the Hamiltonian is identically zero on an optimal trajectory, for a nonlinear time-invariant system
which is to be transferred from an initial state x0 to a final state xf and minimize the performance index
Z Tf
J= g(x(t), u(t))dt
T0
16. A gardener wants to grow plants of a fixed height of 2m in a fixed window of time say [0 , 1] , the growth can
be represented by the dynamic equation:
ẋ = 1 + u
x = Growth state. u = Artificial light input at night.
The cost of lights is given by:
1
u2
Z
J= dt
0 2
Find the optimal light input.
17. Attempt to determine the optimal control law for transferring the system
from an arbitrary initial state to the origin in minimum time with |u(t)| ≤ 1.
18. Consider the following system
(a) Find the unconstrained control, in closed loop form, which minimizes the functional
Z T
1.5x2 (t) + 0.5u2 (t) dt
J=
0
u∗ (t) = F x∗ (t)
3
19. (a) Find the optimal control law for transferring the system
from an arbitrary initial state to the origin in minimum time. The admissible controls are constrained by
|u(t)| ≤ 1.
(b) Generalize the results of part 19a to determine the time-optimal-control law for the system
(a) Determine the costate equations and the required boundary conditions.
(b) Investigate the possibility of singular control intervals.