ELEC 9732 HW1 Solution Notes Solution Notes

Instructions • Given
1 due in Moodle, Tuesday October 29, 4pm       
2 Signed School Cover Sheet attached ẋ1 a b x1 ax1 + bx2
= =
3 PDF only - no microsoft word docs. ẋ2 c d x2 cx1 + dx2

4 Follow the Homework Rules. As in Lec4 reduce this to

Computer
5 output : no discussion ⇒ no marks. ẍ1 = (a + d)ẋ1 + (bc − ad)x1
Analytical
6 results : no working ⇒ no marks.
Taking Laplace transforms gives
7 ♦ means you can use Matlab; else not.
8 No copying except from 9732 lectures. ∆(s)x̄1 (s) = Cs+D where ∆(s) = s2 −(a+d)s−(bc−ad)

for some constants C,D. But check that ∆(s) is the

Q1 (10) Linear System. characteristic polynomial of A and so by assumption
For the two-dimensional linear system ẋ = Ax, it has roots σ ± jω. By inverse Laplace transforms it
consider the case where A has complex eigenvalues follows that
σ ± jω. By converting the system to polar coordi-
nates prove that the phase plane plot has the form x1 (t) = αeσt cos(ωt) + βeσt sin(ωt)
of a spiral. Discuss both stable and unstable cases. x2 (t) = Aeσt cos(ωt) + Beσt sin(ωt)

for some constants α, β, A, B. So we can write

     
x1 z1 β α
= M where M =
x2 z2 BA
   
z1 sin(ωt)
= eσt
z2 cos(ωt)

• We now first analyse the z1 , z2 plane andpthen con-

vert back to the x1 , x2 plane. Let r = z12 + z22 .
σt
and tan(θ) = z1 /z2 . Thus r = e and tan(θ) =
tan(ωt) ⇒ θ = ωt. We find then ṙ = σr and θ̇ = ω
and so
dr σ
= ṙ/θ̇ = r
dθ ω
which is the equation of a spiral. So phase plane
plot in z1 , z2 plane is a spiral. To convert back to
x1 , x2 use the SVD M = U ΛV T ⇒ x = U ΛV T z.
U, V rotate the axes since they are orthogonal while
diagonal Λ scales them so the phase plane plot is still
a spiral.
It is possible to get this result using EVD but te-
dious.

1
 Q2 (20) Phase Plane EPs on [0, π] satisfy x2,e = 0 = 1 − 2sin(x1,e ) ⇒
Consider the following nonlinear system, the so-called sin(x1,e ) = 12 ⇒ x1,e = π6 , 5π 6 . So EPs are:
simplified swing equations for a synchronous machine ( π6 , 0) and ( 5π
6 , 0).
(a rotating mass)  
∂f 0 1
H θ̈ = M − Esin(θ) A(x) = =
∂xT −2cos(x1 ) 0
   
where θ is the rotor angle and the machine is driven 0
√ 1 √0 1
⇒ A π6 = & A 5π =
by a mechanical source supplying a steady power M − 3 0 6 3 0
and is thus generating electrical power E and H is → roots : ±jρ & ± ρ and ρ = 31/4
an inertia term.
Suppose H = 1, M = 1, E = 2. The 1st is a center and the 2nd is a (unstable)
2a Construct a phase plane plot by the method of saddle point. Check perturbation condition.
isoclines.
g(x) = f (x) − A(xe )(x − xe )
    
2b ♦Construct a phase plane plot using matlab. x2 0 1 x1 − x1,e
Discuss the nature of the plot. = −
1 − 2sin(x1 ) −2cos(x1,e ) 0 x2
 
2c Carry out a stability analysis using the Lya- x2 − x2
=
punov indirect method. 1 − 2sin(x1 ) + 2cos(x1,e )(x1 − x1,e )
 
0
=
2∆
Solution Notes ∆ = sin(x1,e ) − sin(x1 ) + cos(x1,e )(x1 − x1,e )

(a) We have θ̈ = 1 − 2sin(θ). Then EPs obey (on Then

[0, π]) 1 − 2sin(θe ) = 0 ⇒ sin(θe ) = 12 ⇒ θe =
π 5π kgk |∆|
6 , 6 . Next ≤ why?
k x − xe k |x1 − x1,e |
dθ̇ dθ̇ ∆
θ̈ = = θ̇ = 1 − 2sin(θ) = | | why?
dt dθ x1 − x1,e
dθ̇ 1 − 2sin(θ) sin(x1 ) − sin(x1,e )
⇒ slope = = = |− + cos(x1,e )|
dθ θ̇ x1 − x1,e
dy 1 − 2sin(x) → | − cos(x1,e ) + cos(x1,e )| = 0
or =
dx y
since x1 → x1,e as k x − x1,e k→ 0 (why?). So
So isoclines satisfy α = (1 − 2sin(x))/y or
perturbation condition holds for either EP.
1
y= (1 − 2sin(x)).
α
and are just scaled versions of each other. Should
show at least 3 cases.
(b) Attached: shows a center and a saddle point.
(c) Convert to SS x1 = θ, x2 = θ̇ ⇒
   
ẋ1 x2
=
ẋ2 1 − 2sin(x1 )

2
 Q3 (20) Stability. (b) The EPs satisfy ẋ1 = 0 = ẋ2 ⇒
Consider the following nonlinear system,
0 = x1,e − x31,e + x2,e & 3x1,e − x2,e = 0
ẋ1 = x1 − x31 + x2 ⇒ x2,e = 3x1,e ⇒ 4x1,e = x31,e
ẋ2 = 3x1 − x2 ⇒ x1,e = 0 or x1,e = ±2
3a Establish existence by showing that a Lipschitz So the EPs are
condition holds in any bounded region,
e.g. {(x1 , x2 ) : |x1 | ≤ R, |x2 | ≤ R}. (0, 0) & (−2, −6) & (2, 6)
3b Find all equilibrium points.
(c) Attached. The given region is small only show-
3c ♦Using matlab, construct a phase plane plot in ing one EP. Plot on larger region shows the other
the region {(x1 , x2 ) : |x1 | ≤ 3, |x2 | ≤ 3} and two as stable nodes
discuss its features. 
1 1

(d) We have (why?) Ao = A0,0 = .
3d Do a linearized stability analysis of one equi- 3 −1
librium point and compare the result with the The eigenvalues satisfy
phase plane plot.
(λ − 1)(λ + 1) − 3 = 0 = λ2 − 4 ⇒ λ = ±2
Solution Notes
so one pole is in the RHP, one in the LHP so an
(a) We have to show unstable system - saddle point.
We must check the perturbation condition.
k f (x) − f (y) k2 ≤ L k x − y k2
g(x) = f (x) − Ao x
There is no MVT for a vector of functions (see
x1 − x31 + x2
   
x1 + x2
appendix B3) so proceed directly. = −
3x1 − x2 3x1 − x2
x1 − x31 + x2 y1 − y13 + y2
   
3
 
−x1
f (x) − f (y) = − =
3x1 − x2 3y1 − y2 0
x1 − y1 + x2 − y2 − (x31 − y13 )
 
kgk |x |3 |x1 |3
=
3(x1 − y1 ) − (x2 − y2 ) ⇒ = p 1 ≤
k x − xe k x21 + x22 |x1 |

Now use (a + b)2 ≤ 2(a2 + b2 ) (why?) = |x1 |2 → 0 as k x − xe k→ 0

and (a + b + c)2 ≤ 4(a2 + b2 + c2 ) (why?)
so the perturbation condition holds.
Then k f (x) − f (y) k22 ≤ ∆2 where
The other EPs have roots satisfying (s + 11)(s +
∆2 = 4(x1 − y1 )2 + 4(x2 − y2 )2 + 4(x31 − y13 )2 1) − 3 = 0 = s2 + 12s + 8 and since both coef-
+ 18(x1 − y1 )2 + 2(x2 − y2 )2 ficients are > 0, Routh-Hurwicz ⇒ two roots in
LHP; so stable nodes.
≤ 22[(x1 − y1 )2 + (x2 − y2 )2 ] + D2
= 22 k x − y k2 +D2
D2 = 4(x31 − y13 )2 = 4[(x1 − y1 )3x2∗ ]2

by the MVT where x∗ lies between x1 , y1 . But

then |x∗ | ≤ R so D2 ≤ 36R4√k x − y k2 . So
Lipschitz is satisfied with L = 22 + 36R4 .

3
 x'=y
y ' = 1 - 2 sin(x)

0
y

-5

-10

-5 0 5 10
x

Figure 1: Q2 Phase Plane Plot

x ' = x - x3 + y
y'=2x-y

0
y

-1

-2

-3

-4

-3 -2 -1 0 1 2 3
x

Figure 2: Q3 Phase Plane Plot