DIRECT DISTURBANCE COMPENSATION
.
13
fedback loops controd also
could
closed depend
For action
the d
on
, No
disturbances directly
acting lit
the be
they measured)
process
on can . o
depend
variable to
manipulated
the
want
We smart
in on
way
a &
dp goal possible
close /DISTURBANCE
to
to Y/Dp
make
is
cre as
: Zero
as lineare
fore
compute
RETECTION) Let's system &
a : dp
· ↳
~
Cals)
H5) H(S)
loop
Y140 doesn't
the change
T(S)
=
= L(s)
1 + 2 o'hy
144)
>
1 )
+
= ...
[d()h(s)
H()
(15)
Y/Dp +
=
L(s) L(S) L(5)
1
1 1 1 )
+ +
+ ...
H(s)/G(s)
if TIDEALCASE)
H(s) (d(s)((s) (d(s)
+ 0
0 < -
= =
=
the distunbance basically
for compensation two problems
solution has
ideal :
frequency
law approximation
usually would
Cd(s) not need
#1 is we a
proper a
it phase far bad delay)
(either poietime
Gls) non-minimum
is
#2 zemos on future
computing dependency
pole
unstable inputs
the AND/aR
matio [d(S)
gives on
an
level dp
of
that
Remember always ununtainty
there approximated
t .
in Ch(s) is
is s
a .
. by
bandwidth
within multiplied
of
Anyway the compensation
imperfect
controllere
the is
any
number
small /
S(s) Y/Dp [d(s)h(s))
1 (H(s) S(S)
((s)
a = 1 + +
: ·
= .
HOW APROXIMATE Cd(s)
DO WE I d
= Be
e
let's G1s)
consider = st
(MP) HI)
# we
, - lcd ** /
5
0 . ·
I > w
H(s)/4(s)
(d(s) 2
have [11 INCTPROPER)
2s)
+ : -odB
-
= =
- I
- o ⑤
CFIS)
lar-frequency E COMPENSATOR
STATIC
· .
approx = +
- for
significant disturbances
if --w
I good
not there high w)
are ** 1800
+25
approximation
second -E -
extend
(S) (we
C -
· can
= . 0 5S
It . of
fast dp
by
the for
pd) Overshoots
approximation at changes
higher changing the
w .
.
1011-s)
let's instead consider the transfer
GIS) get
Sit
H(S)
(NMP)
# 2 5/ We
= .
=
s(1 ,
25)
+
function -112 Instable) that
(d(s) approximate as
can
we :
= ,
law-frequency *
approximation [ -E
(S) STATIC COMPENSATOR
· = < 1 ad
pant -E
to CdmpCdm
compensate
try the (d(s)
NMP
· =
.
= ·
1 actually
it (doesn't
** untain e
1.
approximate past a
work
(s)
C
as -
=
APPROXIMATION ACTUALLY USEFUL
IS
WHEN THE with disturbance rejection
on
Harbandwidth Ya
of
limitations
have strong the
When we
controd) behaviore
[due -
phase 1 -
to non-minimum in
process sensons]
of
moderation the
controd lots of T
Gls) nase
issues on ct
I
,
, then
actually the
measured
dp be compensation
AND can can expl Ts)
H(s) /Its)
(2
= -
be useful Graphical
land it immediatly) example
acts : 10/17)
. 4(s) =
if bad
the ned
then
wed DISTURBANCE (dis)
REJECTION is we
>
additional for distunbance to
want Morraver
take .
into
need
os sensor
we every
one we account ,
distunbance to
quite modelling
sensitive
compensation not mobust)
(ununtainty .
is crimous ,
FREEDOM CONTROLLERS
TWO-DEGREES -
OF
14 . d
If lots of
disturbance
not
consider much
we , con
yoc((s) <41)
set
fast point tracking
and want
(sensors)
noise large to the
follow
need ander
in
very
a w
we e
In
point properly this consider
set case a
can
. we ⑪
PONT
SET COMPENSATION : (p(s)
> d
ACTION DEPEND
DOESN'T
CONTROL
THE ONLY ON THE >
541))380
((s)
BUT y00
FEEDBACK SET
ALSO POINT
CONTROL THE
ON > 0 Y
n
⑦
there
the by
decided
Since setpoint
(DIRECTLY) is us ,
. for
measurament [p(s)
is nase
no We HAVE NOSE
MUCH
EVEN IF WE
WANT
WE
functions
following
consider the
Let's transfer :
= (((41) 225(44)
() +
↑ 143)
/ 1 A(p(s)
=
= =
C(s)G(s)
1 C(s)G(s)
+ 1 +
be problematic
but phase (instability)
solution be
this could non-minimum
Gls)
as
very
can on
law-frequency
to apply approximation
not of
the idea kind
Basically is same
proper :
.
C = function
the
filtim
lar-pas becomes
st
with chosen
F(s) FIS) propera
.
ROCE
PLAY
NaSE DOESN'T WANT
CAN
ALL
ANY AS
AS WE
POINT TRACKING
AT SET FAST
GET
WE
: ((S)
DESIGN
REGARDLESS (S) DISTURBANCE
FEEDBACK
WE FOR
CONTROLLER
OF COMPENSATION
CAN THE
+ .
not
of
lots
with distinbances
Opposite (JET and
much
FILTERING)
PANT
Sunario noise no
,
have
fast
-
Sistemi di controllo - 01(Eng)
-
Sistemi di controllo - 02 (Eng)
-
Sistemi di controllo industriale
-
Sistemi di Controllo