Inter
national
J
our
nal
of
P
o
wer
Electr
onics
and
Dri
v
e
Systems
(IJPEDS)
V
ol.
9,
No.
3,
September
2018,
pp.
1321
–
1329
ISSN:
2088-8694,
DOI:
10.11591/ijpds.v9.i3.pp1321-1329
1321
A
New
Adapti
v
e
Anti-W
indup
Contr
oller
f
or
W
ind
Ener
gy
Con
v
ersion
System
Based
on
PMSG
Ed-dahmani
Chafik,
Mahmoudi
Hassane,
Bak
ouri
Anass,
and
El
Azzaoui
Mar
ouane
Po
wer
Electronics
and
Control
Laboratory
,
Electric
Department,
Mohammed
V
Uni
v
ersity
,
Morocco
Article
Inf
o
Article
history:
Recei
v
ed
Aug
10,
2017
Re
vised
No
v
27,
2017
Accepted
Aug
6,
2018
K
eyw
ord:
Anti-windup
controller
PI
controller
PMSG
WECS
ABSTRA
CT
In
this
paper
,
an
adapti
v
e
anti
-windup
control
strate
gy
for
permanent
magnet
synchronous
generator
dedicated
for
wind
ener
gy
con
v
ersion
systems.
The
proposed
control
has
the
ad-
v
antage
to
suppress
the
performa
nce
deterioration
caused
by
the
o
v
ershooting
phenomenon,
and
optimize
the
controller
g
ains
using
the
particle
sw
arm
optimization
algorithm.
The
scheme
of
the
speed
controller
is
implemented
on
field
orientation
control
in
the
generator
side
con
v
erter
.
A
simulation
of
the
proposed
scheme
is
carried
out
in
SIMULINK-MA
TLAB
in
order
to
e
v
aluate
the
ef
fecti
v
eness
of
the
control
ag
ainst
the
saturation
and
the
parameter
optimization.
Copyright
c
2018
Insitute
of
Advanced
Engineering
and
Science
.
All
rights
r
eserved.
Corresponding
A
uthor:
Ed-dahmani
Chafik
Po
wer
Electronics
and
Control
laboratory
of
Mohammadia
school
of
engineers,
Mohammed
V
Uni
v
ersity
,
Rabat,
Morocco
+212674343036
Email:
chafik.eddahmani@research.emi.ac.ma
1.
INTR
ODUCTION
In
the
last
decades,
there
has
been
a
gro
wing
interest
in
wind
turbines.
Electrical
generators
and
control
strate
gies
should
respond
to
the
needs
of
wind
po
wer
applications.
The
permanent
magnet
synchronous
generator
(PMSG)
and
doubly
fed
induction
generator
(DFIG)
are
widely
used
in
wind
ener
gy
con
v
ersion
systems
(WECS)
because
the
y
of
fer
the
possibility
to
w
ork
with
v
ariable
wind
speed
[1].
Also,
the
y
can
of
fer
an
impro
v
ement
on
production
of
wind
ener
gy
,
and
the
ability
to
achie
v
e
maximum
ener
gy
con
v
ersion
ef
ficienc
y
.
Comparing
to
doubly
fed
induction
generator
(DFIG),
the
PMSG
can
pro
vide
a
high-ef
ficienc
y
and
high
reliability
po
wer
generation,
lo
w
maintenance
required,
and
the
electrical
losses
in
the
rotor
are
eliminated
[2],[3].
Due
to
the
mentioned
adv
antages,
the
PMSG
becomes
an
interesting
solution
for
wind
turbine
applications.
F
or
a
special
architecture
with
a
high
number
of
poles
pairs,
the
PMSG
of
fers
the
possibility
to
eliminate
the
gearbox
system
as
sho
wn
in
Figure
1,
that
allo
ws
a
higher
ef
ficienc
y
.
Se
v
eral
authors
[4],[5]
ha
v
e
discussed
about
the
V
ariable
wind
speed
con
v
ersion
systems
based
on
PMSG.
The
con
v
entional
control
strate
gy
of
PMSG
is
based
on
field
oriented
control
(FOC)
with
PI
controller
[6].
This
controller
is
easy
to
apply
,
and
present
a
good
performance
in
linear
re
gion
[7],[8].
But,
it
suf
fers
from
non-linear
ef
fects
such
as
saturation,
when
the
saturation
is
ne
glected
in
the
design
phase
it
causes
instability
during
closed-loop
where
the
output
system
response
tak
es
a
long
tim
e
to
stabilize
in
the
steady
state.
In
other
w
ords,
this
phenomenon
is
caused
by
the
windup
inte
grator
contained
in
the
PI
controller
,
which
k
eeps
inte
grating
the
tracking
error
e
v
en
if
the
input
is
saturating
[9]–[10].
In
order
to
o
v
ercome
the
windup
phenomenon,
se
v
eral
researches
ha
v
e
proposed
anti-windup
techniques
to
deal
with
input
saturation,
where
some
proposals
in
v
olv
e
a
complicated
design
[11],[12].
Thus,
the
a
v
erage
strate
gy
to
handle
the
inte
grator
windup
by
tuning
the
controller
disre
g
arding
the
saturation
caused
by
the
inte
grator
,
and
then
adds
an
anti-windup
compensator
to
a
v
oid
performance
graduation
[13].
Basically
,
the
classical
anti-windup
strate
gies
come
with
tw
o
dif
ferent
approaches,
namely
,
conditional
inte
gration
and
tracking
back-calculation.
In
this
paper
,
the
speed
control
is
based
on
adapti
v
e
anti-windup
PI
controller
.
The
adaptation
controller
g
ains
are
gotten
with
search
technique
kno
wn
as
P
article
Sw
arm
Optimization
technique
(PSO),
this
technique
w
as
de
v
eloped
by
Eberhart
and
K
ennedy
in
1995
[14],[15].
This
technique
is
an
optimization
tool
based
on
population,
J
ournal
Homepage:
http://iaesjournal.com/online/inde
x.php/IJPEDS
Evaluation Warning : The document was created with Spire.PDF for Python.
1322
ISSN:
2088-8694
and
the
system
is
initialized
with
a
population
of
random
sol
utions
and
can
search
for
optima
by
the
updating
of
generations.
The
PSO
algorithm
has
been
used
in
po
wer
system
for
tuning
control
purpose
in
[16],[17],[18].
The
aim
of
this
paper
is
to
implement,
discuss
and
compare
the
racking
performance
of
adapti
v
e
anti-windup
speed
controller
with
con
v
entional
linear
PI
.
The
proposed
scheme
of
anti-windup
eliminates
the
o
v
ershooting
with
a
simple
structure
e
xisting
in
PI
controllers,
and
guarantees
the
independence
between
the
desired
performance
and
operating
conditions.
The
simulation
is
realized
with
Simulink
in
order
to
v
erify
the
performance
impro
v
ement
com-
paring
to
con
v
entional
PI
controller
.
2.
MODELING
THE
WECS
CONCEPTS
2.1.
W
ind
turbine
model
A
wind
turbine
is
composed
of
man
y
parts
to
con
v
ert
a
kinetic-to-electrical
ener
gy
.
The
mechanical
po
wer
and
torque
deli
v
ered
by
a
wind
turbine
is
gi
v
en
by
(1)
and
(2).
P
t
=
0
:
5
C
p
R
2
v
3
(1)
T
t
=
0
:
5
C
p
R
2
v
3
w
t
(2)
C
p
=
0
:
5
116
i
0
:
4
5
exp
21
i
+
0
:
0068
(3)
W
ith:
1
i
=
1
+0
:
08
0
:
035
1+
3
is
the
air
density
,
R
is
the
blade
radius,
v
and
w
t
are
respecti
v
ely
the
wind
and
turbine
speed,
C
p
is
the
po
wer
coef
ficient,
P
t
is
the
turbine
po
wer
,
and
T
t
is
the
turbine
torque.
The
tip
speed
ratio
(TSR:
)
is
an
important
parameter
in
wind
ener
gy
systems.
It
is
defined
as
the
ratio
of
the
blade
speed
to
the
speed
of
incoming
wind.
PMSG
Grid
RL
fi
l
t
e
r
AC-DC-AC Converters
Figure
1.
WECS
based
on
direct
dri
v
e
PMSG.
0
20
40
60
80
100
120
w
t
( rad.s
-1
)
0
500
1000
1500
2000
2500
P
t
(W)
12 m/s
13.5 m/s
11.5 m/s
10.5 m/s
9 m/s
Figure
2.
T
urbine
po
wer
e
v
olution
curv
e.
2.2.
PMSG
Model
Similarly
to
the
induction
generator
(IG),
the
construction
of
the
stator
in
PMSG
is
essentially
the
same.
On
the
other
side,
the
rotor
magnetic
flux
is
constant
and
generated
by
permanent
m
agnets
[19].
Also,
depending
on
magnets
architecture,
the
PMSG
can
be
classified
into
surf
ace-mounted
and
inset
PM
generators.
The
mathematical
model
of
PMSG
in
synchronous
rotating
dq-reference
is
gi
v
en
by
the
follo
wing
equations.
(
V
sd
=
R
s
:i
sd
L
d
di
d
dt
+
w
r
L
q
:i
sq
V
sq
=
R
s
:i
sq
L
q
di
q
dt
w
r
L
d
:i
sd
+
w
r
f
(4)
T
e
=
3
P
2
((
L
d
L
q
)
i
sd
:i
sq
+
f
:i
sq
)
(5)
J
dw
m
dt
=
T
t
T
e
B
w
m
(6)
Where,
R
s
is
the
stator
resistance,
L
dq
are
the
inductances
in
dq-reference,
f
is
the
PM-flux,
P
is
the
pole
pair
number
,
w
r
and
w
m
are
electrical
and
mechanical
speed,
V
dq
,
I
dq
are
the
stator
v
oltages
and
currents
components
in
dq-reference,
T
e
and
T
t
are
respecti
v
ely
the
electromagnetic
and
turbine
torque,
J
and
B
are
respecti
v
ely
the
equi
v
alent
system
inertia
and
viscous
damping.
IJPEDS
V
ol.
9,
No.
3,
September
2018:
1321
–
1329
Evaluation Warning : The document was created with Spire.PDF for Python.
IJPEDS
ISSN:
2088-8694
1323
The
equations
(4)
represent
the
electrical
beha
vior
of
the
PMSG
which
gi
v
es
the
possibility
to
control
the
dq-currents
components.
Also,
the
speed
controller
is
based
on
(6).
It
should
be
noted
that
the
electromagnetic
torque
T
e
may
be
controlled
directly
by
the
quadrature
current
component
in
case
of
surf
ace
mounted
PM
or
zero
direct
current
control
(ZDC).
Then,
the
e
xpression
of
T
e
is
gi
v
en
by
(7).
T
e
=
3
P
2
f
:i
sq
=
K
t
:i
sq
(7)
3.
WECS
CONTR
OL
STRA
TEGIES
3.1.
W
ind
turbine
contr
ol
The
purpose
by
turbine
control
is
to
produce
a
maximum
po
wer
,
this
is
achie
v
ed
for
a
particular
v
alue
of
po
wer
coef
ficient,
called
C
p
max
,
the
maximal
po
wer
coef
ficient
is
set
to
0.48,
and
it
is
gi
v
en
for
a
specified
TSR
opt
=
8
:
1
with
pitch
angle
=
0
.
The
maximum
po
wer
operation
can
be
achie
v
ed
with
optimal
torque
control
according
to
(9).
opt
=
R
:w
m
opt
v
(8)
T
t
opt
=
0
:
5
C
p
max
R
5
w
2
t
opt
opt
(9)
Also,
to
guarantee
the
safety
w
orking
of
the
turbine
ag
ainst
strong
wind,
a
pitch
angle
controlle
r
is
imple-
mented
in
F
igure
4.
The
w
orking
principle
of
this
controller
is
by
comparing
the
generated
po
wer
P
t
with
the
n
om
inal
po
wer
,
then
it
generates
a
pitch
angle
response.
Eq.2
Eq
.7
+
-
T
e
β
Turbine model
Shaft
MPPT
Eq
.9
Eq
.10
w
t
T
t
w
m
v
C
p
λ
Figure
3.
T
urbine
model
diagram
scheme.
+
-
Pitch angle
contr
oller
P
nom
P
t
β
Figure
4.
pitch
angle
controller
scheme
3.2.
Speed
contr
ol
of
PMSG
The
windup
phenomenon
comes
when
the
output
current
command
of
the
speed
controller
is
limited
to
a
maximum
v
alue,
and
the
inte
gral
component
becomes
v
ery
lar
ge
because
is
not
compatible
with
the
plant
input.
Which
causes
a
lar
ge
o
v
ershoot
and
slo
w
settling
time
in
the
speed
response.
In
this
paper
,
the
speed-loop
controller
has
been
designed
using
an
adapti
v
e
g
ains
anti-windup
algorithm.
It
has
an
adv
antage
to
eliminate
the
o
v
ershooting
phenomenon
during
operation
that
can
deteriorate
the
po
wer
generation
performance.
The
aim
of
anti-windup
control
f
or
nonlinear
system
with
saturating
actuators
is
to
modify
the
control
in
order
to
limit
the
o
v
ershoot.
A
Ne
w
Adaptive
Anti-W
indup
Contr
oller
for
WECS
Based
on
PMSG
(Chafik
ed-dahmani)
Evaluation Warning : The document was created with Spire.PDF for Python.
1324
ISSN:
2088-8694
The
proposed
anti-windup
strate
gy
by
[7]
as
sho
wn
in
fig.5
can
switch
between
the
P
and
PI
modes
according
to
the
w
orking
states.
F
or
the
PI
mode,
it
is
necessary
to
initialize
the
inte
grator
,
and
in
case
of
P
mode
the
initial
v
alue
of
i
sq
i
(0)
is
inserted,
where
the
PI
mode
can
utilize
this
v
alue.
A
lo
w
pass
filter
is
used
to
a
v
oid
an
abrupt
change
of
current
by
loading
the
initial
v
alue
i
sq
i
(0)
in
the
LPF
.
3.3.
Closed
loop
identification
According
to
the
mechanical
equations
of
the
PMSG
(6),(7),
and
for
simplicity
,
the
viscous
damping
B
is
ne
glected,
the
ne
w
e
xpression
of
mechanical
equation
of
WECS
is
gi
v
en
as
follo
w
J
dw
m
dt
=
K
t
:i
sq
T
t
(10)
The
quadrature
stator
current
component
(
i
sq
)
can
be
written
for
the
PI
controller
as
(12).
8
>
<
>
:
i
sq
=
i
sq
i
+
i
sq
p
i
sq
p
=
K
p
(
w
m
r
ef
w
m
)
i
sq
i
=
K
i
s
(
w
m
r
ef
w
m
)
+
i
sq
(0)
s
(11)
Where,
w
m
r
ef
is
the
reference
speed
of
the
PMSG
which
is
the
optimal
speed
deri
v
ed
from
the
MPPT
of
turbine,
i
sq
p
and
i
sq
i
are
the
proportional
and
int
e
gral
components
of
i
sq
,
K
p
and
K
i
are
the
PI
controller
g
ains.
By
implementing
the
Laplace
transform
on
(10)
and
substituting
by
(11),
the
trans
fer
function
of
the
closed
loop
for
the
mechanical
is
e
xpressed
by
(12).
J
(
sw
m
w
m
(0))
=
T
t
s
+
K
t
K
p
+
K
i
s
(
w
m
r
ef
w
m
)
+
i
sq
(0)
s
(12)
w
m
(0)
and
i
sq
(0)
denotes
the
initial
states
v
alues
of
mechanical
speed
and
quadrature
stator
current.
Then,
the
mechanical
speed
w
m
(
s
)
can
e
xpressed
as
a
function
of
the
input
ar
guments
sho
wn
in
(13).
w
m
(
s
)
=
2
6
6
6
4
K
i
:K
t
+
sK
t
:K
p
J
s
2
+
K
t
K
p
:s
+
K
t
K
i
K
t
J
s
2
+
K
t
K
p
:s
+
K
t
K
i
J
s
J
s
2
+
K
t
K
p
:s
+
K
t
K
i
1
J
s
2
+
K
t
K
p
:s
+
K
t
K
i
3
7
7
7
5
T
:
2
6
6
4
W
m
r
ef
i
sq
i
(0)
w
m
(0)
T
t
3
7
7
5
(13)
In
the
steady
stat
e,
according
to
(10)
and
(11),
the
proportional
and
inte
gral
terms
of
quadrature
current
are
e
xpressed
as:
(
i
sq
psteady
=
0
i
sq
isteady
=
T
t
K
t
(14)
When
the
PI
mode
is
acti
v
ated,
the
initial
inte
gral
term
of
i
sq
is
defined
by
(15).
i
sq
i
(0)
=
T
t
K
t
m
(
w
m
r
ef
w
m
(0))
(15)
The
first
term
of
(15)
signify
the
required
current
at
steady
state
according
to
(14),
for
the
second
term
it
is
assigned
to
the
compensation
term
for
the
o
v
ershoot.
W
ith
m
is
the
anti-windup
controller
g
ains.
Substituting
(15)
in
(13).
w
m
(
s
)
=
h
K
t
(
K
i
+
s
(
K
p
m
))
J
s
2
+
K
t
K
p
:s
+
K
t
K
i
K
t
:m
+
J
:s
J
s
2
+
K
t
K
p
:s
+
K
t
K
i
i
:
W
m
r
ef
w
m
(0)
(16)
Then,
(16)
can
be
simplified
during
the
PI
mode
to
the
final
e
xpression
as:
w
m
(
s
)
w
m
(0)
s
=
K
t
(
K
i
+
s
(
K
p
m
))
J
s
2
+
K
t
K
p
:s
+
K
t
K
i
:
w
m
r
ef
w
m
(0)
s
(17)
IJPEDS
V
ol.
9,
No.
3,
September
2018:
1321
–
1329
Evaluation Warning : The document was created with Spire.PDF for Python.
IJPEDS
ISSN:
2088-8694
1325
3.4.
Contr
oller
design
In
order
to
establish
a
good
performance
for
speed
controller
,
the
anti-windup
controller
g
ain
m
should
be
determined
in
the
PI
mode.
According
to
(12),
the
transfer
function
can
be
re
written
as
H
(
s
)
=
K
t
(
K
i
+
s
(
K
p
m
))
J
s
2
+
K
t
K
p
:s
+
K
t
K
i
=
p
1
:p
2
z
:
s
z
(
s
p
1
)(
s
p
2
)
(18)
Where
z
is
the
zero,
and
p
1
;
2
are
the
poles
of
transfer
function.
8
<
:
z
=
K
i
K
p
m
p
1
;
2
=
K
t
:K
p
p
(
K
t
:K
p
)
2
4
J
:K
t
:K
i
2
J
(19)
+
-
+
+
m
-
r
ef
w
∆
w
p
K
s
i
K
m
w
1
/K
t
e
-
r
ef
T
i
sq
i
sq
−
r
ef
w
c
w
c
+
s
LP filter
i
sq
−
i
(0)
PI mode
P mode
i
sq
−
i
i
sq
−
p
<
ma
x
I
Figure
5.
Diagram
scheme
of
the
proposed
anti-windup
speed
controller
.
+
_
+
_
Anti-
w
indup
speed controller
P
MS
G
d
q
abc
d
q
abc
Cur
rent
control
ler
i
dq
+
_
Gener
ator Side c
ontrol
T
t
C
GSC
i
s
d
-
ref
DC-Bus
i
sd
i
sq
w
m
w
m
-
ref
v
d
-
ref
v
q
-
ref
e
d
e
q
w
t
Figure
6.
Global
structure
of
the
generator
side
based
on
the
proposed
control
for
the
mechanical
speed.
As
mentioned
in
[7]
and
[9],
the
pole
is
a
function
of
PI
g
ains,
and
the
zero-location
determined
by
the
anti-
windup
and
PI
g
ains.
In
order
to
simplify
the
transfer
function
to
firs
t
order
as
sho
wn
in
(21),
it
is
assumed
that
z
=
p
1
in
such
a
w
ay
j
p
1
j
<
j
p
2
j
,
which
represents
a
first
order
lo
w
pass
filter
without
saturation.
In
that
case,
the
anti-windup
g
ain
is
gi
v
en
by
(20).
If
the
g
ain
is
smaller
than
the
specified
v
alue,
a
higher
o
v
ershoot
is
pro
vided.
Else,
a
lar
ge
v
alue
can
result
a
slo
w
response.
m
=
K
p
+
K
i
p
1
(20)
H
(
s
)
=
p
2
p
2
s
(21)
The
ne
xt
step
is
to
determine
the
initial
v
alue
of
the
input
ar
guments.
In
transition
state
between
the
P
mode
to
PI
mode,
the
initial
inte
gral
term
of
q-axis
current
component
is
gi
v
en
as
follo
w:
i
sq
i
(0)
=
I
max
K
p
(
w
m
r
ef
w
m
(0))
(22)
Where
I
max
is
the
maxim
um
limited
current.
Also,
according
to
(10),
the
turbine
torque
can
be
e
xpressed
by
in
steady
state.
Using
(15)
and
(22),
the
ne
w
e
xpression
of
i
sq
i
(0)
is
gi
v
en
by
i
sq
i
(0)
=
K
p
:i
sq
isteady
+
m:I
max
K
p
m
(23)
When
the
P
mode
is
selected,
the
initial
inte
gral
current
i
sq
i
(0)
is
generated
by
the
lo
w-pass
filter
.
The
initial
mechanical
speed
w
m
(0)
can
be
e
xpressed
as
follo
w
w
m
(0)
=
w
m
r
ef
+
i
sq
isteady
+
I
max
K
p
m
(24)
3.5.
P
article
Swarm
Optimization
In
the
pre
vious
part,
the
controller
has
been
designed.
The
anti-windup
controller
parameters
g
ains,
as
K
p
,
K
i
and
m
determine
the
performance
system.
The
selection
of
parameters
is
a
task
that
can
be
dif
ficult.
In
order
to
guarantee
a
f
ast-dynamic
response
and
optimal
control,
PSO
comes
to
solv
e
the
problem
of
parameters
estimation.
PSO
deri
v
ed
from
soci
al,
psychologic
al
theory
.
It
imitates
the
natural
process
of
group
communication
to
share
A
Ne
w
Adaptive
Anti-W
indup
Contr
oller
for
WECS
Based
on
PMSG
(Chafik
ed-dahmani)
Evaluation Warning : The document was created with Spire.PDF for Python.
1326
ISSN:
2088-8694
indi
vidual
e
xperience
flocking,
migrating,
or
hunting.
Basically
,
it
searches
for
the
optimal
solution
from
a
population
of
mo
ving
particles.
In
PSO,
starting
with
a
randomly
initialized
population
called
a
sw
arm,
each
member
called
particle
flies
through
the
searching
space,
where
is
positioned
by
x
i
v
ector
,
e
v
aluating
the
fitness,
and
remember
the
best
position
x
g
best
on
which
it
has
the
best
fitness.
This
information
is
shared
by
all
particles
and
adjust
their
positions
x
i
,
and
v
elocities
v
i
according
to
the
information.
The
v
elocity
adjustment
is
based
on
the
historical
beha
viors
of
the
particles
themselv
es
and
their
companions.
The
particles
tend
to
fly
better
to
best
the
positions
[17],[18].
The
v
elocity
and
current
position
respecti
v
ely
of
e
v
ery
particle
are
e
v
aluated
by
(25)
and
(26).
v
i
(
t
+
1)
=
w
:v
i
(
t
)
+
a
[
r
1
(
x
pbest
x
i
(
t
))
+
r
2
(
x
g
best
x
i
(
t
))]
(25)
x
i
(
t
+
1)
=
x
i
(
t
)
+
v
i
(
t
)
(26)
Where,
t
is
time
step,
w
the
inertia
weight
f
actor
,
a
acceleration
constant,
r
1
;
r
2
are
random
functions
in
the
range
of
[0,1],
x
i
the
position
of
i
th
particle,
x
pbest
the
best
pre
vi
ous
position
of
i
th
particle,
x
g
best
the
position
of
best
particle
among
the
entire
population,
and
v
i
the
v
elocity
for
the
i
th
particle.
The
adapti
v
e
weighted
PSO
has
been
proposed
in
(27)
to
impro
v
e
the
reaching
capability
.
a
=
a
0
+
t
N
t
(27)
W
ith
N
t
indicate
the
iterations
number
,
t
is
the
current
step,
and
a
0
is
a
constant
in
[0.5,1].
It
should
be
noted
that
the
inertia
weight
changes
at
e
v
ery
step
by
(28).
w
=
w
0
+
r
3
(1
w
0
)
(28)
W
ith,
w
0
is
a
positi
v
e
constant
chosen
in
[0.5,1],
and
r
3
is
a
random
function
in
the
range
of
[0,1].
4.
SIMULA
TION
RESUL
TS
T
o
v
erify
the
ef
fecti
v
eness
of
the
PI
anti-windup
speed
controller
,
a
simulation
of
the
proposed
scheme
is
car
-
ried
out
in
Simulink.
The
PMSG
and
turbine
data
are
listed
in
T
able
1.In
order
to
e
v
aluate
the
controller
performance
in
e
xtreme
cases,
a
step
change
is
applied
for
the
speed
reference
and
load
torque.
Figure
7
sho
ws
a
comparison
of
the
tracking
performance
of
mechanical
speed
responses
by
the
con
v
entional
PI
and
anti-windup
controllers
with
dif
ferent
iteration
numbers
for
a
step
changing
speed
response
70
r
ad=s
!
157
r
ad=s
!
120
r
ad=s
.
By
comparing
t
he
speed
responses,
in
the
adapti
v
e
anti-windup
controller
,
the
saturation
input
is
limited
which
can
guarantee
a
better
stability
and
high
tracking
performance.
Also,
compared
to
the
con
v
entional
PI
controller
,
the
PSO
impro
v
e
the
controller
performance,
by
selecting
the
optimal
g
ains.
W
ith
the
proposed
method,
the
steady-state
is
quickly
established
for
an
optimal
v
alue
of
the
iterations
number
.Therefore,
the
anti-windup
controller
pro
vides
the
optimal
dynamic
perfor
-
mance
in
term
of
con
v
er
gence,
sat
uration,
and
rob
ustness
compared
to
con
v
entional
PI
controller
.
The
simulation
of
generator
side
con
v
erter
for
the
WECS
is
based
on
the
wind
speed
profile
of
Figure
8,
it
should
be
noted
that
the
nominal
speed
of
the
wind
is
chosen
v
nom
=
12
m:s
1
.
From
Figure
9(a)-9(b),
the
po
wer
coef
ficient
and
tip
speed
ratio-TSR
are
maintained
at
their
optimal
v
alues
by
the
MPPT
control.
The
pitch
angle
controller
is
acti
v
ated
when
the
wind
speed
e
xceeds
the
nominal
speed.
Then,
the
po
wer
coef
ficient
and
tip
speed
ratio
are
decreased
in
order
to
k
ept
the
e
xtracted
po
wer
at
the
nominal
v
alue.
Figure
9(c)
represents
the
pitch
angle
controller
response
for
v
ariable
wind
speed.
The
mechanical
turbine
po
wer
is
il
lustrated
in
Figure
9(d)
changed
according
to
the
wind
speed
v
ariation,
also
it
is
maintained
in
nominal
state
when
the
pitch
angle
controller
is
acti
v
ated.
IJPEDS
V
ol.
9,
No.
3,
September
2018:
1321
–
1329
Evaluation Warning : The document was created with Spire.PDF for Python.
IJPEDS
ISSN:
2088-8694
1327
Figure
7.
Comparison
Simulation
responses
for
con
v
en-
tional
PI
speed-controller
and
the
speed
anti-windup
con-
troller
with
dif
ferent
iterations
number
N
t
0
2
4
6
8
10
12
14
16
18
20
Time (s)
2
4
6
8
10
12
14
v (m.s
-1
)
Wind speed
Nominal speed
Figure
8.
wind
speed
profile.
0
2
4
6
8
10
12
14
16
18
20
Tim
e (
s)
0
0.1
0.2
0.3
0.4
0.5
C
p
(a)
Po
wer
coef
ficient
re
sponse
0
2
4
6
8
10
12
14
16
18
20
Tim
e (
s)
0
5
10
15
(b)
T
ip
speed
ratio
response
0
2
4
6
8
10
12
14
16
18
20
Tim
e (
s)
0
1
2
3
4
(c)
Pitch
angle
response
0
2
4
6
8
10
12
14
16
18
20
Ti
me (
s)
0
500
1000
1500
2000
P
t
(W)
(d)
T
urbine
po
wer
response
Figure
9.
T
urbine
dynamic
performance
using
the
MPPT
and
pitch
angle
controllers
T
able
1.
T
urbine
and
PMSG
data
Nominal
po
wer
P
n
1
:
7
k
W
T
urbine
radius
R
1
:
04
m
Air
density
1
:
22
k
g
=m
3
Gearbox
g
ain
G
1.7
Equi
v
alent
system
inertia
J
0
:
35
N
:m:r
ad
1
:s
2
Maximum
po
wer
coef
ficient
C
p
max
0.48
Optimal
speed
ratio
opt
8.1
Nominal
current
I
nom
5
A
Stator
resistance
R
s
2
:
7
Stator
inductance
L
d;q
3
:
1
mH
PM
flux
f
0
:
341
W
b
Pole
pairs
number
P
4
Nominal
speed
w
nom
157
:
1
r
ad:s
1
Nominal
frequenc
y
f
r
100
H
z
Figure
10(a)
sho
ws
the
high
tracking
performance
of
mechanical
speed
response
of
the
propo
s
ed
controller
under
a
v
ariable
turbine
speed,
it
should
be
noted
that
the
refence
speed
is
gi
v
en
by
the
MPPT
bloc
controller
.
Also,
the
mechanical
speed
response
is
stable
and
tracks
the
reference
v
alue
by
using
the
selected
v
alue
of
the
PSO
algorithm.
A
Ne
w
Adaptive
Anti-W
indup
Contr
oller
for
WECS
Based
on
PMSG
(Chafik
ed-dahmani)
Evaluation Warning : The document was created with Spire.PDF for Python.
1328
ISSN:
2088-8694
In
Fi
gure
10(b),
the
electromagnetic
torque
is
identical
to
the
reference
v
alue.
When,
the
pitch
control
is
acti
v
ated
the
PMSG
speed
and
electromagnetic
torque
are
k
ept
at
the
nominal
v
alues,
which
implies
that
the
e
xtracted
po
wer
is
maximal.
Figure
10(c)
sho
ws
current
responses
in
dq
frame.
The
FOC
with
ZDC
is
applied,
where
the
reference
current
component
of
d-axis
is
set
to
zero
(
i
sd
r
ef
=
0)
as
sho
wn
in
Figure
6,
and
the
quadrature
current
is
propor
-
tional
to
the
turbine
torque
as
mentioned
in
(7).
F
or
the
three-
ph
a
se
stator
current
res
pon
s
e
is
sho
wn
in
Figure
10(d).
Where,
the
current
amplitude
and
frequenc
y
are
proportionals
respecti
v
ely
to
elec
tromagnetic
torque
T
e
and
generator
v
elocity
w
m
.
0
2
4
6
8
10
12
14
16
18
20
Ti
me
(s)
0
50
100
150
w
t
,
w
m
,
w
m-opt
(rad.s
-
1
)
optimal speed Wm-
op
t
actua
l speed
W
m
turbine speed Wt
2
3
4
100
120
140
(a)
Mechani
cal
speed
response
of
PMSG
0
2
4
6
8
10
12
14
16
18
20
Time
(s
)
0
5
10
15
20
T
t
,
T
e
,T
e-re
f
(N.m)
Te
Te-re
f
Tt
0
1
2
6
8
10
(b)
Electromagnetic
torque
response
0
2
4
6
8
10
12
14
16
18
20
Ti
me (
s)
-2
0
2
4
6
8
10
I
dq
(A)
Id
Iq
(c)
Curre
nt
response
in
dq
plan
0
2
4
6
8
10
12
14
16
18
20
Tim
e (
s)
-10
-5
0
5
10
I
abc
(A)
14
14.0
2
1
4.04
-5
0
5
(d)
Stator
current
response
Figure
10.
PMSG
dynamic
performance
with
the
anti-windup
speed
controller
5.
CONCLUSION
The
adapti
v
e
anti-windup
w
as
proposed
in
this
paper
to
replace
the
con
v
entional
PI
controller
for
the
speed
controller
in
the
direct
dri
v
e
PMSG
field
oriented
control.
The
initial
v
alues
of
the
inte
grator
current
and
mechanical
speed
are
determined.
The
PSO
algorithm
is
used
to
estimate
the
optimal
parameters
of
the
proposed
controller
,
which
gi
v
es
a
high
tracking
and
dynamic
performance,
and
f
ast
response
with
least
o
v
ershoot.
The
anti-windup
controller
is
designed
to
gi
v
e
a
best
tracki
ng
speed
comparing
to
the
con
v
entional
linear
PI
controllers.
The
proposed
control
is
implemented
for
the
speed-loop.
The
simulation
results
confirm
the
ef
fecti
v
eness
of
the
anti-windup
controller
re
g
arding
the
saturation
phenomenon
and
f
ast
responses.
REFERENCES
[1]
I.
Boldea,
V
ariable
Speed
Gener
ator
s
,
No
v
.
2005.
[2]
A.
Bak
ouri,
H.
Mahmoudi,
and
A.
Abbou,
“Intelligent
Control
for
Doubly
Fed
Induction
Generator
Connected
to
the
Electrical
Netw
ork,
”
International
J
ournal
of
P
ower
Electr
onics
and
Drive
Systems
(IJPEDS)
,
v
ol.
7,
no.
3,
pp.
688–700,
Sep.
2016.
[3]
M.
E.
Azzaoui,
H.
Mahmoudi,
and
K.
Boudaraia,
“Backstepping
Control
of
W
ind
and
Photo
v
oltaic
Hybrid
Rene
w
able
Ener
gy
System,
”
International
J
ournal
of
P
ower
Electr
onics
and
Drive
S
y
stems
(IJPEDS)
,
v
ol.
7,
no.
3,
pp.
677–687,
Sep.
2016.
[4]
H.
Polinder
,
F
.
F
.
A.
v
.
d.
Pijl,
G.
J.
d.
V
ilder
,
and
P
.
J.
T
a
vner
,
“Compar
ison
of
direct-dri
v
e
and
geared
generator
concepts
for
wind
turbines,
”
IEEE
T
r
ansactions
on
Ener
gy
Con
ver
sion
,
v
ol.
21,
no.
3,
pp.
725–733,
Sep.
2006.
[5]
M.
Chinchilla,
S.
Arnaltes,
and
J.
C.
Bur
gos,
“Control
of
permanent-magnet
generators
applied
to
v
ariable-speed
wind-ener
gy
systems
connected
to
the
grid,
”
IEEE
T
r
ansactions
on
Ener
gy
Con
ver
sion
,
v
ol.
21,
no.
1,
pp.
130–
135,
Mar
.
2006.
[6]
J.
Liang
and
B.
Whitby
,
“Field
Oriented
Control
of
a
Permanent
Magnet
Synchronous
Generator
for
use
in
a
V
ariable
Speed
T
idal
Stream
T
urbine,
”
in
Univer
sities’
P
ower
Engineering
Confer
ence
(UPEC),
Pr
oceedings
of
2011
46th
International
,
Sep.
2011,
pp.
1–6.
IJPEDS
V
ol.
9,
No.
3,
September
2018:
1321
–
1329
Evaluation Warning : The document was created with Spire.PDF for Python.
IJPEDS
ISSN:
2088-8694
1329
[7]
J.
W
.
Choi
and
S.
C.
Lee,
“
Antiwindup
Strate
gy
for
PI-T
ype
Speed
Controller,
”
IEEE
T
r
ansactions
on
Industrial
Electr
onics
,
v
ol.
56,
no.
6,
pp.
2039–2046,
Jun.
2009.
[8]
H.
B.
Shin
and
J.
G.
P
ark,
“
Anti-W
indup
PID
Controller
W
ith
Inte
gral
Stat
e
Predictor
for
V
ariable-Speed
Motor
Dri
v
es,
”
IEEE
T
r
ansactions
on
Industrial
Electr
onics
,
v
ol.
59,
no.
3,
pp.
1509–1516,
Mar
.
2012.
[9]
S.
T
arbouriech
and
M.
T
urner
,
“
Anti-windup
design:
an
o
v
ervie
w
of
some
recent
adv
ances
and
open
problems,
”
IET
Contr
ol
Theory
Applications
,
v
ol.
3,
no.
1,
pp.
1–19,
Jan.
2009.
[10]
R.
J.
W
ai,
J.
D.
Lee,
and
K.
L.
Chuang,
“Real-T
im
e
PID
Control
Strate
gy
for
Magle
v
T
ransportation
System
via
P
article
Sw
arm
Optimization,
”
IEEE
T
r
ansactions
on
Industrial
Electr
onics
,
v
ol.
58,
no.
2,
pp.
629–646,
Feb
.
2011.
[11]
F
.
C.
Ferreira,
T
.
R.
Nascimento,
M.
F
.
Santos,
N.
F
.
S.
Bem,
and
V
.
C.
Reis,
“
Anti
wind-up
techniques
applied
to
real
tank
le
v
el
system
performed
by
PI
controllers,
”
in
2016
20th
International
Confer
ence
on
System
Theory,
Contr
ol
and
Computing
(ICSTCC)
,
Oct.
2016,
pp.
263–268.
[12]
L.
Meng
and
M.
Li,
“
A
ne
w
antiwindup
pi
controller
for
direct
torque
control
system,
”
TELK
OMNIKA
Indonesian
J
ournal
of
Electrical
Engineering
,
v
ol.
12,
no.
7,
pp.
5268–5274,
2014.
[13]
A.
V
isioli,
“Modified
anti-windup
scheme
for
PID
controllers,
”
IEE
Pr
oceedings
-
Contr
ol
Theory
and
Applica-
tions
,
v
ol.
150,
no.
1,
pp.
49–54,
Jan.
2003.
[14]
R.
Eberhart
and
J.
K
ennedy
,
“
A
ne
w
optimizer
using
particle
sw
arm
theory
,
”
in
,
Pr
oceedings
of
the
S
i
xth
Inter
-
national
Symposium
on
Micr
o
Mac
hine
and
Human
Science
,
1995.
MHS
’95
,
Oct.
1995,
pp.
39–43.
[15]
Y
.
Shi
and
R.
Eberhart,
“
A
modified
particle
sw
arm
optimizer
,
”
in
1998
IEEE
International
Confer
ence
on
Evo-
lutionary
Computation
Pr
oceedings.
IEEE
W
orld
Congr
ess
on
Computational
Intellig
ence
(Cat.
No.98TH8360)
,
May
1998,
pp.
69–73.
[16]
Y
.
d.
V
alle,
G.
K.
V
enayag
amoorth
y
,
S.
Mohaghe
ghi,
J.
C.
Hernandez,
and
R.
G.
Harle
y
,
“P
article
Sw
arm
Op-
timization:
Basic
Concepts,
V
ariants
and
Applications
in
Po
wer
Systems,
”
IEEE
T
r
ansactions
on
Evolutionary
Computation
,
v
ol.
12,
no.
2,
pp.
171–195,
Apr
.
2008.
[17]
W
.
Qiao,
G.
K.
V
enayag
amoorth
y
,
and
R.
G.
Harle
y
,
“Design
of
Optimal
PI
Controllers
for
Doubly
Fed
Induction
Generators
Dri
v
en
by
W
ind
T
urbines
Using
P
article
Sw
arm
Optimization,
”
in
The
2006
IEEE
International
J
oint
Confer
ence
on
Neur
al
Network
Pr
oceedings
,
2006,
pp.
1982–1987.
[18]
M.
Y
ang,
X.
W
ang,
and
K.
Zheng,
“
Adapti
v
e
backstepping
controller
design
for
permanent
magnet
synchronous
motor
,
”
in
2010
8th
W
orld
Congr
ess
on
Intellig
ent
Contr
ol
and
A
utomation
,
Jul.
2010,
pp.
4968–4972.
[19]
C.
Ed-dahmani,
H.
Mahmoudi,
and
M.
Elazzaoui,
“Direct
torque
control
of
permanent
magnet
synchronous
motors
in
MA
TLAB/SIMULINK,
”
in
2016
International
Confer
ence
on
Electrical
and
Information
T
ec
hnolo
gies
(ICEIT)
,
May
2016,
pp.
452–457.
A
Ne
w
Adaptive
Anti-W
indup
Contr
oller
for
WECS
Based
on
PMSG
(Chafik
ed-dahmani)
Evaluation Warning : The document was created with Spire.PDF for Python.