Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
9
, No
.
6
,
Decem
ber
201
9
, p
p.
4758
~
4766
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v9
i
6
.
pp475
8
-
47
66
4758
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Bac
ks
tep
pin
g
n
onlin
ea
r
c
ontrol to
ma
ximize en
ergy
ca
ptu
re
in a v
ari
able
sp
eed wind
t
urbin
e
Ouadi
a
El
M
aguiri
1
,
Driss
Mouss
aif
2
, El
ala
mi
Smm
a
3
,
Farc
hi A
bdel
maj
id
4
,
Az
iz
A
khi
at
e
5
1
,
2,3,4
IMM
II
-
lab
,
Facul
t
y
of
Sci
en
ce
and Technolo
g
y
,
Univer
si
t
y
H
assan
1
st
,
Moroc
co
5
RLSEISE
-
la
b
,
Univer
sit
y
of
Ha
ss
an
II,
ENSA
M
,
Moroc
co
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ma
r
1
, 2
01
9
Re
vised
Jun
15
, 201
9
Accepte
d
J
un
27
, 201
9
W
e
are
conside
ri
ng
the
proble
m
of
m
axi
m
u
m
po
wer
point
tra
ck
i
ng
MPPT
in
wind
ene
rg
y
con
ver
sion
s
y
st
em
(W
ECS).
The
p
a
per
proposes
a
n
ew
cont
ro
l
strat
eg
y
to
m
axi
m
iz
e
the
wind
ae
rod
y
n
amic
e
ner
g
y
c
apt
ur
ed
in
var
ia
bl
e
spee
d
wind
turbine
with
a
sepa
ra
te
l
y
exc
i
te
d
DC
-
G
ene
rat
or
and
tr
ansform
ed
to
the
b
at
t
er
y
thr
ough
a
cont
ro
lle
d
DC
-
DC
conve
rte
r.
The
propos
ed
strate
g
y
cont
rols
the
stip
spee
d
rat
io
vi
a
the
rotor
angul
ar
spee
d
to
an
optim
um
point
at
wich
the
power
coe
ffi
ci
e
nt
is
m
axi
m
al
.
The
c
ontrol
ler
is
designe
d
usin
g
the
b
ac
kstepp
ing
technique.
A
fo
rm
al
anal
y
s
is
ba
sed
on
l
y
a
punov
stabilit
y
is
deve
lop
ed
to
d
esc
ribe
the
con
trol
s
y
st
em
per
form
anc
es.
In
addi
ti
on
to
cl
osed
-
loop
g
lo
bal
as
y
m
p
tot
i
c
stabi
lit
y
,
it
is
prove
d
that
th
e
cont
rol
le
r
ac
tu
al
l
y
m
eets
t
he
MP
PT
req
uire
m
e
nt.
Th
e
abo
ve
result
s
ar
e
co
nfirmed
b
y
sim
ula
ti
ons.
Ke
yw
or
d
s
:
Ba
ckstep
ping
c
on
t
ro
l
DC
-
DC con
vert
er
DC gene
rato
r
Ly
apun
ov stab
il
ity
MPPT
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Ou
a
dia E
l
M
a
guiri
,
Dep
a
rtm
ent o
f El
ect
rical
an
d
Me
chan
ic
al
E
nginee
rin
g,
Faculty
of S
ci
e
nce a
nd Tec
hnology,
Un
i
ver
sit
y Ha
s
san 1
st
,
Sett
at
-
Mor
occ
o
.
Em
a
il
:
Ou
adia
_elm
agu
iri@y
ahoo.fr
1.
INTROD
U
CTION
In
isolat
ed
places
wh
e
re
no
el
ect
ric
gr
id
is
avail
able,
wind
tur
bine
s
gen
e
rators
(
W
T
G)
an
d
photov
oltai
c
(P
V)
a
rr
ay
s
are
us
e
d
to
pro
vid
el
ect
rici
ty
.
Ho
wev
e
r
interm
ittent
char
act
eris
ti
c
of
so
la
r
an
d
wind
energy
create
s
a
need
f
or
e
nergy
bac
kup.
Ba
tt
ery
charg
in
g
is
an
i
nteresti
ng
al
te
r
na
ti
ve
beca
us
e
of
it
s
si
m
plici
t
y and
reli
abili
ty
[1
-
2].
As
sho
wn
in
F
ig
ure
1,
batte
r
y
char
gi
ng
a
ppli
cat
ion
of
te
n
us
e
a
ge
ner
at
or
that
conve
rts
wind
tur
bine
powe
r
to
outp
ut
volt
age
an
d
a
DC
-
DC
co
nverter
on
it
s
va
r
iou
s
to
polo
gie
s
Buck
,B
oost,
Buck
-
Bo
os
t
[
3]
and
resona
nt
DC/D
C
conver
te
r
[
4]
.
Gen
e
rall
y
the
DC
ge
ne
rato
r
is
a
go
od
c
hoic
e
f
or
a
(
WT
G)
as
it
is
sim
ply
to
con
t
ro
l
an
d
it
s
dynam
ic
char
act
erist
ic
s
are
ver
y
good
[
5].
On
a
no
t
her
ha
nd,
there
is
a
con
si
der
a
ble
interest
in
us
in
g
va
riable
sp
eed
wind
tur
bin
es
.
I
nd
ee
d,
these
ca
n
be
dr
ive
n
co
ns
t
antly
near
to
the
opti
m
u
m
tip
-
s
pee
d
rati
o
th
rou
gh
tur
bin
e
r
otor
s
peed
c
on
t
ro
l
a
s
s
how
n
i
n
Fig
ure
1
.
S
pecifi
cal
ly
,
ro
t
or
s
pe
e
d
m
us
t
f
ollo
w
wind
-
s
pee
d
va
riat
ion
s
in
l
ow
an
d
m
od
erate
vel
ociti
es
in
orde
r
t
o
m
axi
m
iz
e
aero
dy
nam
ic
eff
ic
ie
ncy.
This sc
hem
e is
know
n
as m
ax
i
m
u
m
p
ow
e
r p
oin
t t
rac
king
(
MPPT)
[
6].
The
Win
d
ge
ne
rators
power
pro
du
ct
io
n
ca
n
be
m
echan
ic
al
ly
con
tr
olled
by
changin
g
t
he
blade
pitch
ang
le
[
7].
H
oweve
r,
(
W
T
G
)
of
s
pecial
const
ru
ct
io
n
are
r
equ
i
red,
w
hich
is
no
t
the
usu
al
case,
especi
al
ly
in
sm
a
ll
-
siz
e stand
-
al
on
e
w
i
nd e
nergy c
onversi
on syst
em
s (
WECS).
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Backstep
ping
nonline
ar co
ntr
ol to m
aximize
ener
gy
c
ap
t
ur
e
in
a
v
ar
ia
ble s
peed wi
nd
.
...
(
Oua
dia
El Ma
gu
iri
)
4759
Figure
1.
Bl
oc
k diag
ram
o
f
th
e
pro
posed
sys
tem
A
c
omm
on
ly
wind
tu
rb
i
ne
con
t
ro
l
m
et
hods
incl
ude
cl
a
ssica
l
te
chn
iq
ues
[8
-
10]
,
w
hich
util
iz
e
a
li
near
iz
ed
wi
nd
t
urbine
syst
e
m
m
od
el
.
A
nothe
r
com
m
on
con
t
ro
l
m
et
ho
d
is
fu
ll
sta
te
fe
edb
ac
k
[11]
w
hich
is
sensiti
ve
to
e
r
rors
in
m
od
el
ing
a
nd
m
easur
em
ents.
In
[
12]
,
an
a
da
ptiv
e
sli
din
g
m
ode
sp
ee
d
co
ntr
ol
is
pro
po
se
d.
Fuzz
y
log
ic
co
ntr
ol
[13]
an
d
ne
ur
al
networks
[
14
]
ha
ve
bee
n
in
ve
sti
gated
to
reduc
e
the
uncertai
nti
es
faced
by
cl
assic
al
con
tr
ol
m
et
ho
ds.
I
n
[
15
]
a
MPP
T
a
lgorit
hm
is
us
ed
f
or
a
pe
rm
anen
t
m
agn
et
ic
synchro
ne
generato
r
us
in
g
gradie
nt
app
r
ox
im
at
ion
.
As
the p
owe
r
coeffic
ie
nt
cp
is
diff
ic
ult
to
ob
ta
in
and
is
diff
e
re
nt
fo
r
e
ver
y
tu
r
bin
e,
a
n
ob
se
r
ver
for
the
est
i
m
ation
of
po
wer
c
oeffici
en
ts
in
a
W
ECS
wh
e
re
a sepa
ratel
y ex
ci
te
d
ge
ner
at
o
r
is u
se
d
i
n [16].
In
this
pa
per
,
a
new
c
ontrol
strat
egy
for
ut
il
iz
ation
of
WECS
in
batte
ry
chargin
g
a
pp
li
cat
io
n
is
pro
po
se
d
i
n
or
der
t
o
ob
ta
in
m
axi
m
u
m
power
point
trac
king
(MP
PT).
A
no
nlinear
c
on
t
ro
ll
er
is
de
velo
ped
us
in
g
the
bac
kst
epp
i
ng
te
c
hniqu
e.
The
c
ontrolle
r
desig
n
is
base
d
on
a
no
nlinear
m
od
el
descr
i
bing
the
wind
tur
bin
e,
the
D
C
ge
ner
at
or,
th
e
boost
co
nver
te
r,
a
nd
the
bat
te
ry.
It
is
f
or
m
al
ly
sh
own
tha
t,
besi
des
cl
os
e
d
lo
op
asym
pto
ti
c sta
bili
ty
, th
e n
on
l
inear contr
oller
actual
ly
m
eet
s
the MPPT r
eq
uir
em
ent. Th
e p
ape
r
is organ
i
zed as
fo
ll
ows:
t
he
c
ontr
olled
syst
e
m
is
m
od
el
le
d
and
gi
ve
n
a
sta
te
sp
ace
re
pr
es
entat
ion
in
Sec
ti
on
2,
the
co
nt
ro
ll
er
desig
ne
d
in
S
ect
ion
3
w
here
it
s
per
f
or
m
ances
are
t
heor
et
ic
al
ly
analy
s
ed,
the
c
ontr
oller
perf
or
m
ances
are
furthe
r
il
lustrat
ed
in
Secti
on
4 t
hro
ugh n
um
e
rical
si
m
ulati
on
s, a
c
on
cl
us
io
n
a
nd a
ref
e
rence
li
st end the
pap
e
r.
2.
SY
STE
M MO
DELL
ING
The rot
or po
w
er
of
t
he win
d
t
urbine
P
w
is give
n by [
17
-
18
]
,
P
w
=
1
2
cp
(
λ
)
ρπ
R
2
ω
w
in
3
(1)
w
he
re
ρ
is
the
ai
r
de
ns
it
y,
R
is
t
he
ra
diu
s
of
t
he
swep
t
a
rea
of
the
tu
rb
i
ne
r
otor,
cp
is
the
power
c
oeffici
en
t
that
is
functi
on
of
t
he
ti
p
-
s
pe
ed
rati
o
λ
fo
r
a
fixe
d
blade
pitch
an
gle,
ω
win
is
the
wi
nd
s
peed.
The
ti
p
sp
ee
d
rati
o
is
def
i
ned as:
λ
=
R
ω
m
ω
win
(2)
w
he
re
ω
m
is
the
wind
tu
r
bi
ne
r
otor
sp
ee
d.
Figure
2
sho
ws
a
ty
pical
cur
ve
of
cp
versus
λ
.
F
or
a
par
ti
cula
r
curve
it
is
poss
ible
to
obta
in
a
poly
no
m
ia
l
app
r
oxim
a
ti
on
f
or
the
cp
as
a
func
ti
on
of
λ
.
The
m
axi
m
u
m
valu
e
of
c
p
,
that
is
c
pm
a
x
=
0
.
407
is
achieve
d
f
or
λ
opt
=
8
.
07
.
This
pa
rtic
ular
va
lue
res
ults
in
the
point
of
op
tim
a
l
eff
ic
ie
ncy
whe
re th
e
w
i
nd turbine ca
ptures t
he
m
axi
m
u
m
p
ow
e
r.
L
c
Gea
rbox
v
a
DC
Gen
e
rato
r
i
b
E
b
R
b
i
L
μ
u
C
i
a
Ba
tt
ery
-
Con
tr
ol
un
it
+
T
m
J
m
T
e
J
e
ω
e
T
T
p
ω
win
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
6
,
Dece
m
ber
2
01
9
:
4758
-
4766
4760
Figure
2. P
ow
e
r
c
oeffici
ent
ve
rsu
s
the
ti
p
s
pe
ed rat
io
T
he p
olyno
m
ial ap
pr
ox
im
at
io
n for
can
be
e
xpresse
d
as
(
)
=
∑
4
=
0
(3)
wh
e
re
the
c
oeffici
ents
have
a
nu
m
erical
va
lues
c
orres
pondin
g
to
cha
rac
te
risti
c
of
us
e
d
WECS.
T
he
orde
r
po
ly
nom
ia
l
is
ta
ken
e
qual
to
=
4
.
Since
involv
es
a
relat
io
n
be
tween
t
he
ro
t
or
sp
ee
d
a
nd
t
he
wind
s
pee
d,
it
is cle
ar th
at
for
a certai
n
wi
nd spee
d, the
re
is a roto
r
s
pee
d
at
witc
h
t
he m
axi
m
u
m
p
ower
po
i
nt
is
rea
ched.
The
m
echan
ic
a
l dynam
ic
o
f
th
e
W
ECS
can
be desc
ribe
d by
the foll
owin
g
s
et
o
f
equati
ons
[15
-
16]
.
̇
+
=
−
̇
+
=
−
=
(4)
w
he
re
J
e
is
the
total
inerti
a
of
DC
ge
ner
at
or,
B
e
is
the
coeffic
ie
nt
of
visc
ous
fr
ic
ti
on.
T
e
is
the
el
ect
ro
m
agn
et
ic
tor
qu
e
.
T
m
is
the
r
otor
t
orq
ue
a
t
the
tur
bin
e
,
J
m
is
the
m
o
m
ent
of
ine
rtia
,
B
m
is
the
f
rict
ion
al
const
ant
of
tur
bin
e
a
nd
is
the
a
ngular
velocit
y
of
the
DC
-
Ge
near
t
or
ro
t
or.
No
te
t
ha
t
no
to
rsion
r
el
at
ed
losses
a
re
consi
der
e
d he
r
e. Furt
her
m
or
e
, th
e t
ran
sm
issio
n i
s ass
um
ed
ideal
.
T
he
t
ran
s
m
issi
on
g
ea
r ra
ti
o
is de
fine
d
a
s
=
(5)
Using
(4)
-
(
5)
we get
the m
echan
ic
al
e
quat
io
n of t
he WECS
:
̇
+
=
−
(6)
wh
e
re
=
+
2
=
+
2
(7)
In
a
bove
e
qu
at
ion
s
,
den
otes t
he
scal
ed
s
um
of
the
ro
t
or
a
nd g
e
ner
at
or
i
ne
rtia
s an
d
the scal
ed
su
m
of
ro
t
or
a
nd
ge
ner
at
or
c
oeffi
ci
ent
of
visco
us
f
rict
ion.
Fi
nally
,
the
W
E
CS
m
od
el
is
obta
ined
by
co
m
bin
ing
the d
y
nam
ic
s o
f
the
D
C
ge
nerat
or
with t
hat of the
turbine
a
nd is
giv
e
n by:
=
−
−
=
1
2
(
)
2
3
−
−
(8)
w
he
re
L
a
an
d
are
res
pecti
vely
the
sta
tor
wi
nd
i
ng
in
duct
a
nce
a
nd
resist
ance.
is
the
induced
em
f
const
ant,
the
f
ie
ld curre
nt a
nd
arm
at
ur
e w
i
nd
i
ng volt
age
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Backstep
ping
nonline
ar co
ntr
ol to m
aximize
ener
gy
c
ap
t
ur
e
in
a
v
ar
ia
ble s
peed wi
nd
.
...
(
Oua
dia
El Ma
gu
iri
)
4761
Let
us
int
rod
uc
e
the
sta
te
va
r
ia
bles
1
=
an
d
2
=
.
T
hen
(
8)
yi
el
ds
the
f
ollow
i
ng
s
ta
te
sp
ace
represe
ntati
on
of the c
om
bin
at
ion
‘turbin
e
–
DC G
e
ne
rato
r’:
1
=
1
2
(
)
2
3
1
−
2
−
1
2
=
1
−
2
−
(9)
Applyi
ng
well
known
ave
ra
ging
te
ch
niqu
e
[3
-
4],
one
obta
ins
the
f
ollow
i
ng
ave
rag
e
m
od
el
f
or
the Bo
os
t
DC
-
DC c
onver
te
r
a
s sho
wn in
F
ig
ur
e
1
:
3
=
−
1
4
+
1
2
4
=
−
(
1
−
)
[
4
+
]
+
1
3
(10)
w
he
re
3
x
and
4
x
de
no
te
res
pecti
ve
ly
the
(av
e
ra
ge
)
volt
age
an
d
current
i
nput
;
is
the
a
ver
a
ge
valu
e
(over
c
utti
ng
per
i
od
s
)
of
the
bi
nar
y
co
ntr
ol
.
Re
cal
l
that
the
co
ntin
uous
sign
al
,
usual
l
y
cal
le
d
duty
r
at
io,
s
ta
nd
s
a
s
the
c
on
t
ro
l
in
put
of
the
syst
e
m
.
The
ob
j
ect
ive
is
to
achiev
e
th
e
MPPT
re
quirem
ent
by
act
i
ng
on
the
duty
rati
o.
The
sta
te
eq
ua
ti
on
s
ob
ta
ine
d
up
t
o
no
w
co
nst
it
ute
a
sta
te
-
sp
ace
re
presen
ta
ti
on
of
the
w
ho
le
syst
e
m
includ
i
ng the
WECS
com
bin
ed wit
h
the
DC
-
DC c
onve
rter a
nd th
e
b
at
te
ry:
̇
1
=
1
(
)
−
1
2
−
2
1
−
3
2
̇
2
=
4
1
−
5
2
−
6
3
̇
3
=
−
7
4
+
7
2
̇
4
=
−
(
1
−
)
[
4
+
]
+
1
3
(11)
w
he
re
1
=
3
2
,
2
=
,
3
=
,
4
=
,
5
=
,
6
=
1
,
7
=
1
3.
CONTR
OLL
ER D
E
SIG
N AND
STABIL
ITY
ANALY
S
IS
To faci
li
ta
te
the contr
ol d
e
vel
oppm
ent p
r
oce
ss,
the
foll
owin
g
ass
um
ption
s
are c
on
si
der
e
d:
A1)
al
l t
he
syst
e
m
p
aram
et
ers
are
known
and
const
ants.
A2)
al
l st
at
es
-
s
pace
(
=
1
.
.
5
)
an
d
th
e
wind s
pee
d
(
)
ar
e m
easur
able.
A2)
t
he win
d
s
peed
(
)
is co
ns
ta
nt or sl
ow
ly
ti
m
e v
aryi
ng (
i
.e
̇
(
)
≅
0
)
The
co
ntr
ol
ob
j
ect
ive
is
to
en
force
the
s
peed
of
the
wind
tur
bin
e
t
o
track
it
s
ref
ere
nce
tr
ajecto
ry
,
=
−
1
.
Th
e
m
axim
u
m
po
we
r
po
i
nt
is
r
eached
w
he
n
=
.
Fo
ll
owin
g
the b
ac
ks
te
pp
i
ng tech
nique
[
19
-
21
]
the
cont
ro
ll
er is
d
esi
gned
in
fo
ur
ste
ps:
St
ep
1.
le
t us
intr
oduce t
he
s
pe
ed
trac
king e
r
ror:
1
=
1
−
2
=
1
−
1
∗
(12)
In v
ie
w of
(
3)
and (
11)
t
he
a
bove
er
ror u
nd
e
rgoes
t
he
(13)
:
̇
1
=
1
∑
−
1
2
−
2
1
−
3
2
−
̇
1
∗
5
=
0
(13)
In
(13)
t
he
qu
antit
y
=
−
2
2
sta
nds
up
as
a
(v
i
rtu
al
)
con
t
ro
l
in
put
for
the
1
-
dy
nam
ic
s.
Let
∗
de
no
te
the stabil
iz
ing
functi
on (
ye
t t
o be
determ
ined) asso
ci
at
ed
t
o
. I
t i
s ea
sil
y see
n from
(
13)
t
hat if
=
∗
with:
∗
=
−
1
1
−
1
∑
−
1
2
+
3
1
+
̇
1
∗
5
=
0
(14)
The
n
on
e
wou
ld
get
̇
1
=
−
1
1
with
1
>
0
is
a
desig
n
pa
ram
et
er.
This
would
cl
early
ens
ur
e
asym
pto
ti
c stabil
it
y of
(1
3) wi
th r
es
pect the
Ly
apun
ov fun
c
ti
on
:
1
=
0
.
5
1
2
(15)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
6
,
Dece
m
ber
2
01
9
:
4758
-
4766
4762
I
n ef
fect, the
ti
m
e
-
der
ivati
ve of
1
would t
he
n be:
̇
1
=
1
̇
1
−
1
1
2
<
0
(16)
As
=
−
2
2
,
is
just
a
vi
rtual
co
ntr
ol
i
nput,
one
ca
n
no
t
set
=
∗
.
Ne
vert
heless,
t
he
a
bove
ex
pressi
on
of
∗
retai
ne
d
as
a fi
rst stabil
iz
at
ion f
un
ct
io
n
a
nd a
n
e
w
e
rror i
s in
tro
du
ce
d:
2
=
−
∗
(17)
Using
(14)
-
(17
),
it
foll
ows
fro
m
(
13
)
t
hat the
1
-
dynam
ic
s u
nder
goes the
(
18)
:
̇
1
=
−
1
1
+
2
(18)
St
ep 2
.
T
he
obj
ect
iv
e
now
is
to
en
force
the error
v
aria
ble
to
va
nis
h.
(
1
,
2
)
.To
th
is
end,
le
t
us
fi
rst
determ
ine
the d
y
nam
ic
s o
f
2
:
̇
2
=
−
2
̇
2
−
̇
∗
(19)
Ba
sed on t
he
a
ssu
m
ption
A2
;
(19) to
gethe
r wit
h (11) an
d (
14)
im
plies:
̇
2
=
−
1
1
−
1
∑
(
−
1
)
−
2
̇
2
+
2
1
−
2
(
4
1
−
5
2
−
6
3
)
5
=
0
(20)
wh
e
re i
n view
of (5)
̇
≅
̇
1
(21)
T
he
qua
ntit
y
2
=
2
6
3
ta
nds
as
vi
rtual
c
ontrol
in
(
20),
le
t
2
∗
de
no
te
s
the
c
orr
esp
ondi
ng
sta
bili
zi
ng
fun
ct
ion
ass
ociat
e
d
to
2
, it i
s clea
r
that if
2
=
2
∗
with:
2
∗
=
−
2
2
−
1
−
1
∑
(
−
1
)
(
−
2
)
̇
2
+
(
2
4
−
2
)
1
−
2
5
2
5
=
0
(22)
The
n
one
w
ould
get
̇
2
=
−
2
2
−
1
with
2
>
0
is
a
desi
gn
par
am
et
er.
This
would
cl
early
ens
ure
asym
pto
ti
c stabil
it
y of
(
1
,
2
)
erro
rs
w
it
h res
pect th
e Lya
puno
v
f
unct
ion
:
2
=
0
.
5
1
2
+
0
.
5
2
2
(23)
A
s
2
=
2
6
3
is
just
a
vi
rtual
c
on
t
ro
l
input,
one
can
no
t
set
2
=
2
∗
.
Ne
vert
heless,
t
he
a
bove
e
xpress
i
on
of
2
∗
retai
ned
as
a first sta
bili
zat
ion
functi
on a
nd the t
hire
d
e
r
ror
is i
ntrod
uce
d:
3
=
2
6
3
−
2
∗
(24)
U
sin
g (22)
-
(24
),
it
foll
ows
fro
m
(
20
)
t
hat the
2
-
dynam
ic
s u
nder
goes the
(
25)
:
̇
2
=
−
2
2
−
1
+
3
(25)
St
ep
3.
Let
us
inv
est
igate
t
he beha
vior
of the
erro
r
3
, in
view
of (2
4), tim
e d
erivati
ve o
f
3
gi
ves:
̇
3
=
2
6
̇
3
−
̇
2
∗
(26)
F
ro
m
(
22
)
,it
is
r
ea
dily
seen
t
ha
t
̇
2
∗
=
−
2
̇
2
−
̇
1
+
2
4
̇
1
−
2
5
̇
2
−
2
̇
1
−
1
∑
(
−
1
)
(
−
2
)
(
(
−
2
)
(
−
3
)
̇
+
̈
)
2
5
=
0
(27)
U
sin
g (11) a
nd (22) it
foll
ows
f
r
om
(
26)
t
hat:
̇
3
=
2
6
(
−
7
4
+
7
2
)
+
2
̇
2
+
̇
1
−
(
2
4
−
2
)
̇
1
+
2
5
̇
2
+
1
∑
(
−
1
)
(
−
2
)
5
=
0
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Backstep
ping
nonline
ar co
ntr
ol to m
aximize
ener
gy
c
ap
t
ur
e
in
a
v
ar
ia
ble s
peed wi
nd
.
...
(
Oua
dia
El Ma
gu
iri
)
4763
(
(
−
2
)
(
−
3
)
̇
+
̈
)
2
(28)
I
n
(
28),
the
f
ollow
i
ng
qu
a
nt
it
y
3
=
−
2
6
7
4
act
s
a
s
virtu
al
co
ntr
ol.
T
o
determ
ine
a
c
orres
ponden
t
sta
bili
zi
ng
fun
ct
ion
3
∗
, let us
conside
r
the
qua
drat
ic
augm
ente
d
Ly
a
punov f
unct
ion can
di
date:
3
=
0
.
5
1
2
+
0
.
5
2
2
+
0
.
5
3
2
(29)
T
i
m
e
-
der
ivati
ve
of
3
al
ong
t
he t
raj
ect
ory
of
t
he
syst
e
m
(
1
,
2
,
3
)
is gi
ven b
y:
̇
3
=
−
1
1
2
−
2
2
2
+
2
3
+
3
̇
3
(30)
T
he
a
bove
exp
ressio
n of
̇
3
sug
gest the
foll
owing st
abili
zi
ng
functi
on
3
∗
=
−
3
3
−
2
6
7
2
−
2
̇
2
−
̇
1
+
(
2
4
−
2
)
̇
1
−
2
5
̇
2
−
2
−
1
∑
(
−
5
=
0
1
)
(
−
2
)
(
(
−
2
)
(
−
3
)
̇
+
̈
)
2
(31)
wh
e
re
3
>
0
is a
ne
w param
et
er desi
gn. Let
us
i
nt
rodu
ce
the
fo
ll
ow
i
ng er
ror
4
=
3
−
3
∗
(32)
C
om
bin
ing
(31
)
a
nd (32) yi
el
ds
̇
3
=
−
3
3
−
2
+
4
(33)
St
ep
4.
The
obj
ect
ive
no
w
is
to
en
f
or
ce
the
e
rro
r
var
i
ables
(
1
,
2
,
3
,
4
)
to
va
nish
,
to
t
his
e
nd
,
le
t
us
deter
m
ine the
dynam
ic
s o
f
4
. Deri
ving (3
2)
and usi
ng (1
1)
on
e
obtai
ns
:
̇
4
=
2
6
7
(
(
1
−
)
[
4
+
]
+
1
3
)
−
̇
3
∗
(34)
w
ith
̇
3
∗
=
−
3
̇
3
−
2
6
7
̇
2
−
2
̈
2
−
̈
1
+
(
2
4
−
2
)
̈
1
−
2
5
̈
2
−
̇
2
−
1
∑
(
−
1
)
(
−
5
=
0
2
)
(
−
2
)
(
(
−
3
)
(
−
4
)
̇
2
+
(
−
3
)
̈
+
⃛
)
2
−
1
∑
(
−
1
)
(
−
2
)
(
−
3
)
̇
(
(
−
5
=
0
2
)
(
−
3
)
̇
+
̈
)
2
(35)
In
(35), the act
ual control in
put nam
el
y
arises for
the f
i
rst tim
e;
w
e seek
the stabil
iz
at
ion of
the
fu
ll
error sy
ste
m
(
1
,
2
,
3
,
4
)
with
resp
ect
t
o t
he
f
ollo
wing a
ug
m
ented
ly
ap
u
no
v functi
on
cand
i
date:
4
=
3
+
0
.
5
4
2
(36)
It is easi
ly
ch
e
cked that
, t
he
t
i
m
e d
erivati
ve of
4
is given b
y:
̇
4
=
−
1
1
2
−
2
2
2
−
3
3
2
+
3
4
+
4
̇
4
(37)
I
t i
s easil
y che
cked that
t
he
a
bove de
rivati
ve
sug
gests the
foll
ow
i
ng b
ac
kst
epp
in
g
c
ontr
ol law
=
1
−
[
3
+
(
−
4
4
−
3
+
̇
3
∗
)
2
3
7
]
1
(
+
4
)
(38)
w
he
re
4
>
0
is a
ne
w desig
n para
m
et
er; that i
s (37)
bec
om
es:
̇
4
=
−
∑
2
4
=
1
<
0
(39)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
6
,
Dece
m
ber
2
01
9
:
4758
-
4766
4764
The res
ults o
btained u
p
to
no
w
a
re sum
m
arized in
the
f
ollo
wing the
orem
.
Theorem
(Ma
in
resu
lt
s)
.
Co
ns
ide
r
the
syst
e
m
of
fig.1
an
d
desc
ribe
d
by
the
m
od
el
(1
1)
in
cl
os
e
loop
with
the b
ac
ks
te
pp
i
ng contr
oller
(
38); the
n o
ne h
as the
fo
ll
owin
g pro
per
ti
es:
1
-
al
l si
gn
al
s a
re
bounde
d
a
nd t
he
trac
king e
rror
s
(
1
,
2
,
3
,
4
)
van
is
h
asy
m
pto
ti
cally
2
-
the
m
axi
m
iz
a
tio
n o
f
t
he
e
nerg
y ca
ptured
is a
chieve
d.
Pro
of
.
the
(
1
,
2
,
3
,
4
)
er
r
or
-
syst
em
is g
iven
the c
om
pact form
[
1
2
3
4
]
=
[
−
1
−
1
0
0
1
−
2
−
1
0
0
1
−
3
−
1
0
0
1
−
4
]
[
1
2
3
4
]
(40)
it
is read
il
y see
n from
(
39)
tha
t
̇
≤
−
1
‖
[
1
2
3
4
]
‖
2
(41)
w
ith
1
=
{
1
,
2
,
3
,
4
}
(42)
w
he
re
‖
.
‖
denotes
the
Eucli
dian
norm
,
these
ens
ur
e
that
th
e
equ
il
ib
rium
(
1
,
2
,
3
,
4
)
=
(
0
,
0
,
0
,
0
)
of
the syst
em
(
11)
is
globall
y ex
pone
ntial
ly
sta
ble [2
0]. T
he
l
at
te
r
ens
ur
e
the
MPPT
obj
ect
i
ve.
4.
PERFO
R
MANC
ES
EV
AL
UA
TI
ON
The
perform
a
nces
of
t
he
pro
posed
c
on
t
ro
l
desig
n
are
il
lustrat
ed
th
rou
gh
si
m
ulati
on
s.
The
ex
pe
rim
en
ta
l
set
up
,
has
been
sim
ulate
d
in
Ma
tl
ab/Si
m
ul
ink
en
vir
onm
ent.
The
inv
ol
ved
el
em
ents
have
the foll
owin
g
c
har
act
erist
ic
s
a
s sho
wn in
Tab
le
1
:
Table
1.
T
he
c
h
aracte
risti
cs
of the
in
vo
l
ved
el
e
m
ents
Batter
y
Eb=
2
4
V;
R
b
=0
.6
5
Ω
DC
-
DC co
n
verters
L
c
=1
mH;
C=
4
.7m
F;
DC g
en
era
to
r
L
a
=1
1
mH;
R
a
=1
.
2
Ω
;
J
e
=0
.20
8
kg
m²
Be=0
.01
1
kg
m²
;
Ke=
0
.35
3
;
If=0
.15
A.
Air den
sity
25
.
1
W
in
d
tur
b
in
e
J
m
=0
.1kg
m²
;
B
m
=0
.01
5
kg
m
²
;
R=0
.5m
;
opt
=6
;
Tra
n
smis
sio
n
gear
ra
tio
5
.
11
The val
ues
of
(
=
0
.
.
4
)
in (3) are:
0
=
121
×
10
−
4
;
1
=
−
302
×
10
−
4
;
2
=
196
×
10
−
4
;
3
=
−
34
×
10
−
4
;
4
=
2
×
10
−
4
The
c
ontrolle
r
perf
or
m
ances
will
be
e
val
uated
i
n
pr
es
ence
of
(tim
e
-
va
ryi
ng)
wind
velocit
y.
The
wi
nd
sp
ee
d
re
fer
e
nce
as
sh
ow
n
in
F
ig
ure
3
ta
ke
s
a
lo
w,
m
edium
an
d
hi
gh
value
(
equ
al
to
8,
10.
7,
14.
6
and
ste
p
to
12
.5
m
/s
at
t
i
m
es
0,
200,
400
a
nd
500s
res
pe
ct
ively
).
W
it
h
these
val
ues
of
wind
s
pee
ds
,
blo
c
k
op
ti
m
iz
ation
ge
ner
at
es
t
he
optim
al
ro
tor
s
pe
ed
r
efe
ren
ce
sh
ow
n
in
Fig
ure
3.
T
he
i
nd
ic
at
ed
val
ues
of
desig
n
par
am
et
ers
1
=
75
;
2
=
20
;
3
=
18
an
d
4
=
15
hav
e
bee
n
s
el
ect
ed
us
in
g
a
try
-
an
d
-
e
rror
search
m
et
hod
and
prov
e
d
t
o
be
su
it
able.
T
he
con
tr
oller
pe
rfor
m
ances
are
il
lustrate
d
by
Fig
ure
3
an
d
F
ig
ure
4;
they
sh
ow
that
the
m
achi
ne
s
peed
perf
ect
ly
con
ve
rg
e
to
it
s
ref
e
rence
.
The
t
rack
i
ng
q
ualit
y
is
qu
it
e
sat
isfact
or
y
a
s
the
respo
ns
e
ti
m
e
(af
te
r
each
change
in
the
wind
s
peed
)
is
le
ss
than
10s
.
The
pe
rf
ect
MPPT
in
pr
e
s
ence
of
wind
sp
ee
d
ch
ang
e
s
is
s
how
n
i
n
Fig
ure
4
.
T
he
sp
ee
d
rati
o
c
onve
rg
es
t
o
it
s
opti
m
a
l
va
lue
=
8
.
07
f
or
wh
ic
h
the
po
w
er c
oeffici
ent
(
)
=
0
.
407
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Backstep
ping
nonline
ar co
ntr
ol to m
aximize
ener
gy
c
ap
t
ur
e
in
a
v
ar
ia
ble s
peed wi
nd
.
...
(
Oua
dia
El Ma
gu
iri
)
4765
Figure
3.
W
i
nd
sp
ee
d
a
nd tu
rbi
ne
r
otor s
pee
d p
rofil
es
Figure
4.
W
i
nd sp
ee
d rati
o
an
d
powe
r
c
oeffici
ent
5.
CONCL
US
I
O
N
In
t
his
pa
pe
r,
a
ne
w
s
olu
ti
on
to
M
PPT
of
var
ia
ble
s
pee
d
sm
al
l
wind
batte
ry
-
ch
ar
ger
syst
e
m
is
dev
el
op
e
d.
M
P
PT
is
achieve
d
us
in
g
the
nonl
inear
back
ste
ppin
g
co
ntr
oller
(38)
base
d
on
the
syst
e
m
no
nl
inear
m
od
el
(1
1).
T
he
co
ntr
oller
is
prov
e
n
to
yi
el
d
a
glo
ba
ll
y
un
i
form
l
y
bo
unde
d
sta
ble
cl
os
e
d
-
lo
op
syst
e
m
via
ly
apu
no
v
-
base
d
analy
sis.
Si
m
ula
ti
on
res
ults
dem
on
strat
e
the
sat
isfact
ory
perform
ance
of
the
pro
pose
d
con
t
ro
ll
er.
REFERE
NCE
S
[1]
Z.
Ch
en
and
F
.
Bla
abjerg
,
“
W
ind
Ene
rg
y
-
The
W
orld’s
Fastest
Grow
ing
Ene
rg
y
Source
,
”
IE
EE
Powe
r
Elec
tronic
s
Soci
e
ty
N
ewsle
t
t
er
,
vol
.
18
,
pp
.
1
5
-
19
,
2006
.
[2]
D.
C.
Phan
and
T.
H.
Tri
nh
,
“
Maximum
Powe
r
Ex
tracti
on
Method
for
Doubl
y
-
f
ed
Induc
ti
on
Gene
rat
or
W
ind
Turbi
ne
,
”
Inte
rn
ati
onal
Journal of
E
le
c
tric
al
and
Computer
Eng
i
nee
ring (
IJE
C
E)
,
v
ol
.
8
,
pp
.
711
-
722
,
2018
.
[3]
H.
E
.
Fad
il
and
F.
Giri,
“
Ba
ckst
eppi
ng
Based
C
ontrol
of
PW
M
DC
-
DC
Boost
P
ower
Converters
,
”
Pro
ce
ed
ings
o
f
the
I
EE
E
ISIE
'0
7,
V
igo, Spai
n
,
p
p
.
395
-
400
,
200
7
.
[4]
O.
E
.
Maguiri
,
e
t
al
.
,
“
Adapti
ve
cont
rol
of
a
cl
as
s
of
serie
resona
nt
DC
-
DC
conv
ert
er
,
”
IF
AC
Sy
mpos
ium
on
Powe
r
Pl
ants a
nd
Power
Syste
ms
Contr
ol
,
2009
.
[5]
W
.
Li,
e
t
al.
,
“
Resea
rch
on
W
ind
Turbi
ne
E
m
ula
ti
on
base
d
on
DC
Motor
,
”
2nd
IEE
E
C
onf.
on
Industrial
El
e
ct
ronics
and
App
li
ca
ti
ons,
IC
IEA
2007
,
pp
.
25
89
-
2593
,
2007
.
0
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
6
0
0
8
9
10
11
12
13
14
15
t
i
m
e
(
s
)
w
i
n
d
s
p
e
e
d
(
m
/
s
)
0
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
6
0
0
0
50
1
0
0
1
5
0
2
0
0
2
5
0
t
i
m
e
(
s
)
r
o
t
o
r
s
p
e
e
d
(
r
d
/
s
)
o
p
t
i
m
a
l
s
p
e
e
d
a
c
t
u
a
l
s
p
e
e
d
0
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
6
0
0
0
1
2
3
4
5
6
7
8
9
10
t
i
m
e
(
s
)
t
i
p
s
p
e
e
d
r
a
t
i
o
0
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
6
0
0
0
.
1
0
.
1
5
0
.
2
0
.
2
5
0
.
3
0
.
3
5
0
.
4
0
.
4
5
0
.
5
0
.
5
5
0
.
6
t
i
m
e
(
s
)
Cp
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
6
,
Dece
m
ber
2
01
9
:
4758
-
4766
4766
[6]
D.
V.
N.
Ananth
and
G.
V.
N
.
Kum
ar,
“
Ti
p
spe
e
d
rat
io
base
d
MP
PT
al
gorit
hm
a
nd
improved
fi
eld
orie
nt
ed
cont
r
ol
for
ext
r
acting
op
ti
m
al
r
ea
l
power
and
inde
p
ende
n
t
re
ac
t
ive
power
cont
ro
l
for
grid
connect
ed
doub
l
y
fe
d
induction
gene
ra
tor,”
In
te
r
nati
onal Journal
of El
e
ct
ri
cal
an
d
Computer
Eng
ine
ering
,
vol
.
6
,
pp.
1319
-
1331,
2
016.
[7]
F.
Le
sche
r
,
et a
l
.
,
“
Robust Ga
in
Sch
edul
ing
Controll
er
for
Pitc
h
Reg
ula
t
ed
Vari
able S
pee
d
W
ind Tur
bine
,
”
S
tudi
es
in
Informati
cs
a
nd
Control
,
v
ol
.
14,
pp
.
299
-
315
,
2005.
[8]
T.
Knuds
en,
et
al
.
,
“
Com
par
ing
PI
and
robu
st
cont
rol
pitch
co
ntrol
lers
on
a
4
00KW
wind
turbi
ne
b
y
ful
l
sca
le
te
sts
,
”
Depa
r
tment
of
Con
trol E
n
gine
er
ing, Aal
bo
rg
Univer
sit
y
,
A
al
borg,
Denm
ark
,
T
ec
h
.
R
ep. R
-
97
-
4174,
1997
.
[9]
B.
Boukhezz
ar
a
nd
H.
Siguerdi
dj
ane
,
“
Nonlinear
Control
of
Vari
a
ble
-
Speed
W
ind
Turbi
nes
for
Gene
ra
tor
Torqu
e
L
imiti
ng
and
Po
wer
Optimizatio
n
,
”
J. Sol. Energ
y
Eng
.
,
v
ol
.
128
,
p
p.
516,
2006.
[10]
B.
Boukhe
zzar,
et
al
.
,
“
Multi
v
ariabl
e
cont
ro
l
str
a
te
g
y
for
var
i
able
spee
d,
var
ia
bl
e
pit
ch
w
ind
turb
i
nes,
”
Re
n
ewabl
e
Ene
rgy
,
vol
.
32
,
pp.
1273
-
1287
,
2007.
[11]
K.
Stol
and
M.
Bal
as,
“
Full
-
sta
t
e
fee
db
ac
k
con
tr
ol
of
a
var
i
abl
e
-
spee
d
wind
turbine:
A
compari
son
of
per
iodi
c
an
d
consta
nt
Gains
,
”
J. of Sol
ar
Ene
r
gy
Eng
.
,
vol
.
12
3,
pp
.
319
-
326
,
2001.
[12]
A
.
M
era
b
et
,
“
A
dapt
ive
slidi
n
g
m
ode
cont
ro
l
for
wind
en
erg
y
exp
eri
m
en
ta
l
s
y
stem
,”
En
ergie
s
,
vo
l.
11
,
pp.
1
-
14
,
2018
.
[13]
X.
Zha
ng
,
e
t
a
l.
,
“
Fuzz
y
cont
rol
of
var
ia
b
le
spe
e
d
wind
turbi
n
e
,
”
Proc.
6
th
Worl
d
Congress
on
Inte
lligent
Contr
ol
and
Aut
omat
ion
,
pp.
3872
-
3876,
2006.
[14]
F.
Kane
ll
os
and
N.
Hatz
i
arg
y
riou
,
“
A
new
cont
rol
sche
m
e
for
var
i
abl
e
spe
ed
wind
turbi
ne
using
ne
u
ral
ne
tworks
,
”
Proc.
I
EEE
P
ower
Eng. Soc.
Tr
a
ns.
Distrib.
Conf
.
,
New York
,
N
Y
,
2002
,
pp.
360
-
3
65.
[15]
S.
Morim
oto,
et
al.
,
“
Pow
er
m
axi
m
iz
a
ti
on
con
trol
of
var
i
able
-
spee
d
wind
g
en
era
t
ion
s
y
s
te
m
using
per
m
ane
n
t
m
agne
t
s
y
nchr
o
nous ge
ner
a
tor
,
”
Tr
ans.
Inst.
El
e
c.
Eng. of J
apan
,
Part B
,
vol
.
123
-
B,
pp
.
1573
-
15
79,
2003
.
[16]
L
.
D,
K.
B
and
M.
J
.,
“
Esti
m
at
ion
of
the
power
coe
ffi
ci
e
n
t
in
a
wind
conve
rsion
sy
st
em
,
”
Proceedi
ngs
of
the
44t
h
IEE
E
Confe
ren
c
e
on
D
ec
ision
an
d
Control, and t
he
European
Co
ntrol
Conf
ere
nce 2005
Se
vil
le
,
Sp
ain
,
2005
.
[17]
I.
Munte
anu,
e
t a
l.
,
“
Optimal
Con
trol
of
W
ind
En
e
rg
y
S
y
st
ems
,
”
Springer, 2008.
[18]
F.
Senani
,
et
a
l.
,
“
A
Co
m
ple
te
Modeli
ng
and
Con
trol
for
W
ind
Tu
rbine
Based
of
a
Doubl
y
Fed
Ind
uct
ion
Gen
era
to
r
using
Dire
ct
Pow
er
Control
,
”
Inte
rnational
Jou
rnal
of
Powe
r
El
e
ct
ronics
and
Dr
iv
e
Syste
ms
(
IJP
EDS)
,
v
ol.
8,
pp.
1954
-
1962
,
2017.
[19]
J.
Zhou
,
e
t
al
.
,
“
Adapti
ve
b
ac
ks
te
pping
cont
ro
l
of
Separ
a
tel
y
E
xci
t
ed
DC
Moto
r
with
Unc
ertain
ti
es
,
”
I
EE
Proc
.
El
e
ct
.
Pow
er
Ap
pl.
,
vol
.
149
,
pp
.
165
-
172
,
2002
.
[20]
Krs
ti
c
M.
,
e
t al
.
,
Nonli
near
and
a
dapti
v
e
con
trol design
,
John
W
ila
y
&
Sons
,
Inc
,
1995.
[21]
M.
E
.
Azz
aoui,
et
al
.
,
“
Bac
ksteppi
ng
Control
of
wind
and
ph
otovol
taic
h
y
bri
d
Rene
wable
E
ner
g
y
S
y
s
te
m
,
”
Inte
rnational
Jo
urnal
of Powe
r
El
e
ct
ronics
and
Dr
iv
e
Syst
ems (
IJP
EDS)
,
v
ol
.
7
,
pp.
677
-
686
,
20
16.
Evaluation Warning : The document was created with Spire.PDF for Python.