M d li f C ti ti t ModelingofContinuoustimesystem

We have two approaches for modeling the linear

continuous time system: y
1) Transfer function approach
2) State space representation 2) State space representation
Transfer function approach make use of the Laplace
Transform.
Idea
the Laplace transform converts integral and dierential equations into
algebraic equations
this is like phasors, but
applies to general signals, not just sinusoids
handles non-steady-state conditions
allows us to analyze
LCCODEs
complicated circuits with sources, Ls, Rs, and Cs
complicated systems with integrators, dierentiators, gains
The Laplace transform 32
The Laplace transform
well be interested in signals dened for t 0
the Laplace transform of a signal (function) f is the function F = L(f)
dened by
F(s) =
_

0
f(t)e
st
dt
for those s C for which the integral makes sense
F is a complex-valued function of complex numbers
s is called the (complex) frequency variable, with units sec
1
; t is called
the time variable (in sec); st is unitless
for now, we assume f contains no impulses at t = 0
common notation convention: lower case letter denotes signal; capital
letter denotes its Laplace transform, e.g., U denotes L(u), V
in
denotes
L(v
in
), etc.
The Laplace transform 34
Example
lets nd Laplace transform of f(t) = e
t
:
F(s) =
_

0
e
t
e
st
dt =
_

0
e
(1s)t
dt =
1
1 s
e
(1s)t

0
=
1
s 1
provided we can say e
(1s)t
0 as t , which is true for s > 1:

e
(1s)t

e
j(s)t

. .
=1

e
(1s)t

= e
(1s)t
the integral dening F makes sense for all s C with s > 1 (the
region of convergence of F)
but the resulting formula for F makes sense for all s C except s = 1
well ignore these (sometimes important) details and just say that
L(e
t
) =
1
s 1
The Laplace transform 35
More examples
constant: (or unit step) f(t) = 1 (for t 0)
F(s) =
_

0
e
st
dt =
1
s
e
st

0
=
1
s
provided we can say e
st
0 as t , which is true for s > 0 since

e
st

e
j(s)t

. .
=1

e
(s)t

= e
(s)t
the integral dening F makes sense for all s with s > 0
but the resulting formula for F makes sense for all s except s = 0
The Laplace transform 36
sinusoid: rst express f(t) = cos t as
f(t) = (1/2)e
jt
+ (1/2)e
jt
now we can nd F as
F(s) =
_

0
e
st
_
(1/2)e
jt
+ (1/2)e
jt
_
dt
= (1/2)
_

0
e
(s+j)t
dt + (1/2)
_

0
e
(sj)t
dt
= (1/2)
1
s j
+ (1/2)
1
s + j
=
s
s
2
+
2
(valid for s > 0; nal formula OK for s = j)
The Laplace transform 37
powers of t: f(t) = t
n
(n 1)
well integrate by parts, i.e., use
_
b
a
u(t)v

(t) dt = u(t)v(t)

b
a

_
b
a
v(t)u

(t) dt
with u(t) = t
n
, v

(t) = e
st
, a = 0, b =
F(s) =
_

0
t
n
e
st
dt = t
n
_
e
st
s
_

0
+
n
s
_

0
t
n1
e
st
dt
=
n
s
L(t
n1
)
provided t
n
e
st
0 if t , which is true for s > 0
applying the formula recusively, we obtain
F(s) =
n!
s
n+1
valid for s > 0; nal formula OK for all s = 0
The Laplace transform 38
Impulses at t = 0
if f contains impulses at t = 0 we choose to include them in the integral
dening F:
F(s) =
_

0
f(t)e
st
dt
(you can also choose to not include them, but this changes some formulas
well see & use)
example: impulse function, f =
F(s) =
_

0
(t)e
st
dt = e
st

t=0
= 1
similarly for f =
(k)
we have
F(s) =
_

0

(k)
(t)e
st
dt = (1)
k
d
k
dt
k
e
st

t=0
= s
k
e
st

t=0
= s
k
The Laplace transform 39
Linearity
the Laplace transform is linear : if f and g are any signals, and a is any
scalar, we have
L(af) = aF, L(f + g) = F + G
i.e., homogeneity & superposition hold
example:
L
_
3(t) 2e
t
_
= 3L((t)) 2L(e
t
)
= 3
2
s 1
=
3s 5
s 1
The Laplace transform 310
One-to-one property
the Laplace transform is one-to-one: if L(f) = L(g) then f = g
(well, almost; see below)
F determines f
inverse Laplace transform L
1
is well dened
(not easy to show)
example (previous page):
L
1
_
3s 5
s 1
_
= 3(t) 2e
t
in other words, the only function f such that
F(s) =
3s 5
s 1
is f(t) = 3(t) 2e
t
The Laplace transform 311
what almost means: if f and g dier only at a nite number of points
(where there arent impulses) then F = G
examples:
f dened as
f(t) =
_
1 t = 2
0 t = 2
has F = 0
f dened as
f(t) =
_
1/2 t = 0
1 t > 0
has F = 1/s (same as unit step)
The Laplace transform 312
Inverse Laplace transform
in principle we can recover f from F via
f(t) =
1
2j
_
+j
j
F(s)e
st
ds
where is large enough that F(s) is dened for s
surprisingly, this formula isnt really useful!
The Laplace transform 313
Time scaling
dene signal g by g(t) = f(at), where a > 0; then
G(s) = (1/a)F(s/a)
makes sense: times are scaled by a, frequencies by 1/a
lets check:
G(s) =
_

0
f(at)e
st
dt = (1/a)
_

0
f()e
(s/a)
d = (1/a)F(s/a)
where = at
example: L(e
t
) = 1/(s 1) so
L(e
at
) = (1/a)
1
(s/a) 1
=
1
s a
The Laplace transform 314
Exponential scaling
let f be a signal and a a scalar, and dene g(t) = e
at
f(t); then
G(s) = F(s a)
lets check:
G(s) =
_

0
e
st
e
at
f(t) dt =
_

0
e
(sa)t
f(t) dt = F(s a)
example: L(cos t) = s/(s
2
+ 1), and hence
L(e
t
cos t) =
s + 1
(s + 1)
2
+ 1
=
s + 1
s
2
+ 2s + 2
The Laplace transform 315
Time delay
let f be a signal and T > 0; dene the signal g as
g(t) =
_
0 0 t < T
f(t T) t T
(g is f, delayed by T seconds & zero-padded up to T)
PSfrag replacements
t t
t = T
f(t) g(t)
The Laplace transform 316
then we have G(s) = e
sT
F(s)
derivation:
G(s) =
_

0
e
st
g(t) dt =
_

T
e
st
f(t T) dt
=
_

0
e
s(+T)
f() d
= e
sT
F(s)
The Laplace transform 317
example: lets nd the Laplace transform of a rectangular pulse signal
f(t) =
_
1 if a t b
0 otherwise
where 0 < a < b
we can write f as f = f
1
f
2
where
f
1
(t) =
_
1 t a
0 t < a
f
2
(t) =
_
1 t b
0 t < b
i.e., f is a unit step delayed a seconds, minus a unit step delayed b seconds
hence
F(s) = L(f
1
) L(f
2
)
=
e
as
e
bs
s
(can check by direct integration)
The Laplace transform 318
Derivative
if signal f is continuous at t = 0, then
L(f

) = sF(s) f(0)
time-domain dierentiation becomes multiplication by frequency
variable s (as with phasors)
plus a term that includes initial condition (i.e., f(0))
higher-order derivatives: applying derivative formula twice yields
L(f

) = sL(f

) f

(0)
= s(sF(s) f(0)) f

(0)
= s
2
F(s) sf(0) f

(0)
similar formulas hold for L(f
(k)
)
The Laplace transform 319
Integral
let g be the running integral of a signal f, i.e.,
g(t) =
_
t
0
f() d
then
G(s) =
1
s
F(s)
i.e., time-domain integral becomes division by frequency variable s
example: f = , so F(s) = 1; g is the unit step function
G(s) = 1/s
example: f is unit step function, so F(s) = 1/s; g is the unit ramp
function (g(t) = t for t 0),
G(s) = 1/s
2
The Laplace transform 325
derivation of integral formula:
G(s) =
_

t=0
_
_
t
=0
f() d
_
e
st
dt =
_

t=0
_
t
=0
f()e
st
d dt
here we integrate horizontally rst over the triangle 0 t
PSfrag replacements
t

lets switch the order, i.e., integrate vertically rst:

G(s) =
_

=0
_

t=
f()e
st
dt d =
_

=0
f()
_
_

t=
e
st
dt
_
d
=
_

=0
f()(1/s)e
s
d
= F(s)/s
The Laplace transform 326
Multiplication by t
let f be a signal and dene
g(t) = tf(t)
then we have
G(s) = F

(s)
to verify formula, just dierentiate both sides of
F(s) =
_

0
e
st
f(t) dt
with respect to s to get
F

(s) =
_

0
(t)e
st
f(t) dt
The Laplace transform 327
examples
f(t) = e
t
, g(t) = te
t
L(te
t
) =
d
ds
1
s + 1
=
1
(s + 1)
2
f(t) = te
t
, g(t) = t
2
e
t
L(t
2
e
t
) =
d
ds
1
(s + 1)
2
=
2
(s + 1)
3
in general,
L(t
k
e
t
) =
(k 1)!
(s + 1)
k+1
The Laplace transform 328
Convolution
the convolution of signals f and g, denoted h = f g, is the signal
h(t) =
_
t
0
f()g(t ) d
same as h(t) =
_
t
0
f(t )g() d; in other words,
f g = g f
(very great) importance will soon become clear
in terms of Laplace transforms:
H(s) = F(s)G(s)
Laplace transform turns convolution into multiplication
The Laplace transform 329
lets show that L(f g) = F(s)G(s):
H(s) =
_

t=0
e
st
_
_
t
=0
f()g(t ) d
_
dt
=
_

t=0
_
t
=0
e
st
f()g(t ) d dt
where we integrate over the triangle 0 t
change order of integration: H(s) =
_

=0
_

t=
e
st
f()g(t ) dt d
change variable t to t = t ; dt = dt; region of integration becomes
0, t 0
H(s) =
_

=0
_

t=0
e
s(t+)
f()g(t) dt d
=
_
_

=0
e
s
f() d
__
_

t=0
e
st
g(t) dt
_
= F(s)G(s)
The Laplace transform 330
examples
f = , F(s) = 1, gives
H(s) = G(s),
which is consistent with
_
t
0
()g(t )d = g(t)
f(t) = 1, F(s) = e
sT
/s, gives
H(s) = G(s)/s
which is consistent with
h(t) =
_
t
0
g() d
more interesting examples later in the course . . .
The Laplace transform 331
Finding the Laplace transform
you should know the Laplace transforms of some basic signals, e.g.,
unit step (F(s) = 1/s), impulse function (F(s) = 1)
exponential: L(e
at
) = 1/(s a)
sinusoids L(cos t) = s/(s
2
+
2
), L(sint) = /(s
2
+
2
)
these, combined with a table of Laplace transforms and the properties
given above (linearity, scaling, . . . ) will get you pretty far
and of course you can always integrate, using the dening formula
F(s) =
_

0
f(t)e
st
dt . . .
The Laplace transform 332
Patterns
while the details dier, you can see some interesting symmetric patterns
between
the time domain (i.e., signals), and
the frequency domain (i.e., their Laplace transforms)
dierentiation in one domain corresponds to multiplication by the
variable in the other
multiplication by an exponential in one domain corresponds to a shift
(or delay) in the other
well see these patterns (and others) throughout the course
The Laplace transform 333