0% found this document useful (0 votes)
43 views13 pages

Assignment 1

This document summarizes an optimal control problem. It shows that the control u* = 3t is a successful control that minimizes the cost function and satisfies the boundary conditions. Specifically: 1) It tests the initial and terminal conditions for the control u* = 3t and shows they are satisfied. 2) It calculates the cost function J for u* = 3t and shows it equals 1/3, the minimum value. 3) It gives the corresponding trajectory x*(t) = t3/2 + t2 for the successful control u* = 3t.

Uploaded by

Jaco Greeff
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
43 views13 pages

Assignment 1

This document summarizes an optimal control problem. It shows that the control u* = 3t is a successful control that minimizes the cost function and satisfies the boundary conditions. Specifically: 1) It tests the initial and terminal conditions for the control u* = 3t and shows they are satisfied. 2) It calculates the cost function J for u* = 3t and shows it equals 1/3, the minimum value. 3) It gives the corresponding trajectory x*(t) = t3/2 + t2 for the successful control u* = 3t.

Uploaded by

Jaco Greeff
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

APM4809 -

Assignment or 2023/os/os
Unique Code:840978 JW GREEff (21136521)

1) Exercise 1. I

GIVEN State equation: tu


=

...

Initial condition:x(0) 0
=

... 2

Terminal condition:x(1) 1
=
...

cost function:J: S'u2dt


=

...

NowIS** dt =
Stu dt ... by
Eg

x(t)
=>
Stndt
=

...

THEN testing the initial and terminal conditions:

x(0) =

0 fotndt
=

0
=

=> the initial condition in 2


> atisfies $

x)
=

1 S:
=

tudt 66 a
terminal
8, by condition in

Now
showing It
n*- is a successful control:

I Jn
=
=
dt ...
by

=S.(n -st) 6tu


+ -
qt2dt...(n 3t)2 -
u2
=

-
6t at+
2

=
(.(n -
3t)2 dt 6(1) + -

3t3!!...by 0

=S.(u -3H2 dt + 3

of
since the
integral non-negative, the
is min 5 is 3

which can be achieved by


setting u st
=

..n*=st is a successful control


the
corresponding costfunction for u* 3t:
=

I Ju 2dt
=

... by

.* J:(st) 2dt 1!a


=
=

2dt
+
3
=

t10 3
=

the
corresponding trajectory for u* 3t:
=

x(t) 1tudt
=

...
by $

·x*(t) =

I3t2dt +)! =
=

+
3

(t) t the successful


the
is
corresponding trajectoryfor
=

control u* 3t=

f(x) = u*
=

v
+
... assumed to be another successful control

&

e
u st
=

v
+ and &i t(3t v)
=

+
st2 ut
=
+

...byQ

:. the solution of the new state equation:

x(t) (t3t2
=
+
vt dt =

t3 Stv dt
+
t

s(t) satisfies the condition((0)


initial
>
c
=

↳ x(=1 1
=

J!rtdt=> 1!vtdt
+
o...
=

· the cost ofthis control:

3 1!u
=
=
dt

=S. (st v)2 dt


+

=S!at2 6tv
+ v2
+ dt

=J'at2dt Sovdt
+

...
by

3
=
+
f!v2dt
>
3 since
integral non-negative
:.3 x
3 3*and
=

J 3*=)
=

v 0
=

THUSI is
n* the optimal control
Exercise
2) 12

GIVEN State equations.


Sincewith active
...

Initial conditions:(,(0). 0,xz(0) x =

= > 0 ... 2

Terminal conditions:(,(t) =

0, x2(t) 0
=

...

cost function:I =
S"idt t,
=

...

SINCE) the controls are banded as


given by -1 4,
=
It)=1 Yt we know
I of
subset the optimal control set
1 where the control variables only
take their extreme values

·any change in control results in the control switching from one

extreme to another

·
: we consider the 2 extreme values:II as
separate cases

CASE, where 4, (t) =-1:

Nawl ic,
3 with the [C's:(,(0). 0,x(0) x
=

x
=
= > o

xz 1
=

THEN1 1.c2dt=-I. dt => xt) X


=
-

by
t
the IC's

·J.c, dt I!xz(t)
=
dt

=>
x,(t) 3![X
=
-
3dt xt
+
=
-

·
the solution rector:jt(t (ii] [X]
=
=

: the solution vector satisfies the initial conditions and

the terminal conditions:


by

CASE
2 where U, (t) 1
=

: with terminal time t, ct

Now, atsome time to with tactwe switch the control to the


value of+ 1.

ALSO/ic, 2
3 with the IC's:x,(0). xz(0) x >
=
0, 0
=
=

xz
=
1
THEN SE, c2 dt Si,
=

dt => xalt) t- t,
=

...

·(FeatIcecre ...
by

El-titl
=

= -

E -

(t,t -t,
I
=
-

I
+

I(t- t,)2
=

THUs) itis possible to reach the attime to provided


origin
we can solve for t, and ts s.t


E x2(t) X t far u,(t) =
=

1
- -

...

x,(t) xt -
=

②\<(t) t- t, for n,(t) 1


= =

. . .

x,2t) I(t t,)2


=
-

*:X tz t,
-
t =
-
=

2tz =
x +

t,
t=x +
=>
tz 2 +
=

tz -
t, =

0 t2 t,
=
=

I
It 2, +1) x z- X tz tz t,
=

&
+
- + - = -

It
=> -

Ent, aXtz =
(=
(t, -

tz) tz = -

t,(t,
=> -

tz) 2t22
+
2Xt z
=

(
t,(tz -

x) 2tz(X
=
-

(z)
E t,
2
=
①Ein- It

E? En t

C
+ x +

·.xz(t,) t,
=

2tz =

0 tz
= =

It,
-:.x,(ti)
I

It? -

2t,tz x
+ t z2
+
=

It
=> -

t, +
x +
0
=

=>
- E x
+
+
=

x
=> =

=t=2X
3) Exercise 2.2

GIVEN in =c, + txz


3 with x,(0) and x2(0) c2
=

c,
=

i=x2

NOTE here we have a coupled system

2-T1 = = (=2]
=
A := constantmatrix, fundamental
X:= matrix

SINCEl cEC1) Aci (t) =

A
=> =
[di]
ALSO1 the fundamental matrix X has a det Fo

·: find the eigenvalues & of A:


me
eigenvectors

1A-x2) =

0
1 =x)
=
=

==> (1 -
x) 0:X=

1
= with multiplicity 2

for eigenvector : (8] (vic] (8] =

:(* 2] [0] =

=> viz 0
=

:.x, =

[]
For
eigenvector i2: [8][22] [] =

..[*] [8] =

v2z
=>

1
=

:.[i]
.
. (t) ar,et blt,+vz) et
= +

=>

(ii] et((:] =

(-))
+

(x,(t) (at bt 2) et
=> =

xz(t) bet =

the initial
by conditions:x,(0) c,xz(0) =

ca
=

..c, 0
=
md cr b
=

(at ct) et
...[x,(t)
=

czet
...
xz(t) =
4)Exercise 2.3

(a) GIVENI A [00]


=

LET)c rextbe = a trial solution ofi As with =

55=(=2]

Nowl bythe Characteristic Eg:1A-XII 0


=

.('* (x) =

0 =
x2 -
x =

X(x
=> -
1) =

X, 0,xz
=> =

1
=

F2x), an
eigenvector corresponding to X, =

·:[d]() (8] =

n, v,
=

+
0
=

r
=

[i]
=

FIX) is an
eigenvector corresponding to x2 1
=

:. (8-J[i] (8] =

= (-22] (:] =
i
=>
[6]
=

SINCE) we have a linearlyindependenteigenvectors:

x,(t) eti, 2 -i]


= =

xr(t) extvz
=

[]
=

THUSIX Fundamental
=
=

Matrix=['."I
(b) GIVENI A [8p']
=

LET)c rextbe = a trial solution ofi As with =

55=(=2]

Nowl bythe Characteristic Eg:1A-XII 0


=

·(Ppx) =

0
(p
=
-
x)2 0
=

=>
X p =

with multiplicity
2

F2x), an
eigenvector corresponding to X, P
=

·: [8][] [8] =
=>
v,z 0=
i
=

[d]
=

farv, = 1

FIX) is an
eigenvector corresponding to x2 P
=

:. (8!][n]= [i] = vez 1


=

v
=
[9]
=

SINCE) we have a linearlyindependenteigenvectors:

:x,(t) ei,
=

ePt[d] [2- I
=
=

x(t) ext(tv, vz)


=
+
et
=

[t] (t 28] =

THUSIX = fundamental matrix-


=

[etet]
(c) GIVENI A= L 88]
LET =nextbe a trial solution ofin- As with i=(?]
Nowl bythe Characteristic Eg:1A-XII 0
=

·(i) =

0 - -
x3 0
=

X
=>
=0 with multiplicity 3

F2x), an
eigenvector corresponding to X, 0
=

·188:][] (8] (5](:] = => =


= (:] farm

FIX) is an
eigenvector corresponding to x2 0
=

·
18:](*] (8] =
=
(w] (!] =

r
=>
(8]
=

forvai

FIX)

an eigenrectalcorresponding test(i]
is

· 5=
=>
(i] for t

SINCE we have 3 linearlyindependenteigenvectors:

:x,(t) e4i,
=

(!) =

(a(t) tex2 +,
=

(*] rz
+
=

xy(t) texstr,
=

+ +3 (*)
=

THUSIX = fundamental
=

matrix=(o*****]
(d) GIVENI A [8 :g]
=

LET)c rextbe = a trial solution ofi As with =

55=(=2]

Nowl bythe Characteristic Eg:1A-XII =


0

·( qx) -
=

0 - x) q
=
- -
x) =

->
1 0,Xz= =

F2x), an
eigenvector corresponding to X, 0
=

·: (8-][] (8] =

[rz] (8)
=>
=

=>
v,e 0
=

v
=

(0]
=

FIX) is an
eigenvector corresponding to x2 =
-

:. L a
·
b][v] [8] =
=> [21, 22]= [8] v2
=

=-
i
=>

[-,*]
=

SINCE) we have a linearlyindependenteigenvectors:

:.x,(t) eti,
=

[i]
=
-

xa(t) ent =
=

2
=

e qt[-,]

THUSIX = fundamental
=

matrix=[deat) I
(e) GIVEN A 1=
=

LET)c rextbe = a trial solution ofi As with =

55=(=2]
Now the of diagonal matrixis values its
eigenvalues a
equal to on

diagonal.

·. In-numer ofeigenvalues:Xi=PiXi =,...,n

ALSO the eigenvectors of a


diagonal matrix
in M" is simplythe standard
basis ofRM

:. In-number of
eigenvectors as the cols of V with

V Cr,
=
rc ...
[n] with vi =
eigenvectors
=

of A

=diag(1, ...,1) sit vi corresponds to Xi


-
n -
number of Xi=1, ...,
entries

SINCE) we have a linearlyindependenteigenvectors:

:xilt) eiti Vi=1,...,


-

THUSIX:Fundamental Matrix, is diagonal with


X diag (etr.,
=

...,
etron(
5) GIVENI
E Es with <,(0) x and (0) 0 initial
=

co x2
=

...

conditions
... a coupled system

LET)x,(t,) =

0 and xz(t) 0
=

...
terminal conditions

FIx) J(t,x,,xz,u) t function


given
=

... cost

ASSUME) the control is


bang-bang s.t n -
=

a or n b
=

for a,b>0

:.u =

b for ostite and n = -


a for tz<t<t,

RTSI target be reached


can with
single switch

SINCE) X co in the initial state -> C, is initiallynegative att 0 =

and then x, 0
=

at t t,
=

·u will be positive for time tats and


negative for time

t >
t2

⑧ t z ti
for
Now Solving , and x2:
ret I
I
-
I

CASE, for It (0, tn)

Navl J.*2dt=1.* dt

xr(t)
=>
bt)
=

btz
=

THEN 1
·J*, dt =
Stxz(t) dt

x,(t)
=>
(bt zdt
=

btzt)* =

bt2

Buil by the IC's:,

·
the solution rector:jt(t (ii] [X]
=
=

: the solution vector satisfies the initial conditions and

the terminal conditions:


by

You might also like