Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
6, N
o
. 4
,
A
ugu
st
2016
, pp
. 14
06
~
1
411
I
S
SN
: 208
8-8
7
0
8
,
D
O
I
:
10.115
91
/ij
ece.v6
i
4.1
073
5
1
406
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
A Review of Wind Speed Estima
tion f
o
r Wind Turbine Syst
ems
Based on Kalman Filter Technique
M. N
a
j
a
fi Kh
oshr
odi
1
, Mo
ha
mma
d
Ja
nnat
i
2
, Tole
Sutik
no
3
1
Golestan
E
l
ec
tr
ica
l
Power Distr
i
bution Com
p
an
y
2
UTM-PROTON Future Driv
e
Laborator
y
,
Faculty
of
Elec
trical Engin
eering
,
U
n
iver
siti Teknologi Malay
s
ia,
81310 Skudai, Johor Bahru, Malay
s
ia
3
Department of
Electrical Eng
i
n
eering
,
Un
iv
ersitas Ahmad Dahlan, Yog
y
ak
arta, I
ndonesia
Article Info
A
B
STRAC
T
Article histo
r
y:
Received
Ma
r 3, 2016
Rev
i
sed
May 15
, 20
16
Accepted
May 30, 2016
This paper pres
ents a review of
wind speed estimation based
on Kalman
filte
r techn
i
que
applied to wi
nd turbine s
y
st
em
s. Generall
y,
wind speed
m
eas
urem
ent is
perform
ed b
y
an
em
om
eter. Th
e
wind s
p
eed prov
ided b
y
th
e
anem
om
eter is
m
eas
ured at
a s
i
ngle poin
t
of
the
rotor pl
ane
whi
c
h is
not
th
e
accur
a
t
e
wind s
p
eed. Als
o
, us
i
ng anem
om
eter increas
es
the s
y
s
t
em
cos
t
,
m
a
intenan
c
e
,
c
o
m
p
lexit
y
and
reduces
th
e r
e
li
abili
t
y
. F
o
r
the
s
e reas
ons
,
estimation of wind speed is needed for
wind tur
b
ine s
y
st
em
s. In
this paper
,
the several wind speed estim
atio
n m
e
thods based on Kalm
an filter m
e
thod
us
ed for wind
tu
rbine s
y
s
t
em
s
ar
e rev
i
ewed
.
Keyword:
Kalm
an
Filter
Rev
i
ew
Spee
d Estim
ation
Wi
n
d
T
u
r
b
i
n
e
Copyright ©
201
6 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
M. N
a
j
a
f
i
Kh
osh
r
od
i,
Golesta
n
Elect
rical Powe
r
Di
st
ri
but
i
o
n C
o
m
p
any
.
Em
a
il: m
n
aj
afik
ho
sh
rod
i
92
@g
m
a
i
l
.co
m
1.
INTRODUCTION
W
i
n
d
ener
gy
i
s
pl
ay
i
ng a si
gni
fi
can
t ro
le
in
fu
ture en
erg
y
scen
e. About 10% electri
city can b
e
su
pp
lied b
y
t
h
e w
i
nd
en
er
g
y
b
y
th
e year
20
20
.
W
i
n
d
e
n
e
r
gy is
one t
h
e
fastest inc
r
easing and e
n
vironm
ent
fri
en
dl
y
re
ne
w
a
bl
e e
n
er
gy
s
o
urces
.
Wi
n
d
t
u
rb
i
n
es co
nv
ert
th
e k
i
n
e
tic en
erg
y
of the
wi
nd to electrical
energy
[
1
]-[
5
]
.
Gen
e
rally, wind
tu
rb
in
es are categ
o
r
ized
in
to
two
typ
e
s, fix
e
d
and
v
a
riab
le sp
eed
wi
n
d
turb
i
n
es
wh
ich
v
a
riab
l
e
sp
eed
wind tu
rb
i
n
e is mo
re
reliab
l
e.
To
con
t
ro
l the v
a
riab
le sp
eed
wi
n
d
turb
in
es, th
e
measurem
ent of wind spee
d is requ
ire
d
. For this, anem
o
m
eters are place
d to m
easure the wind s
p
eed. W
i
nd
turbine a
n
em
ometer can
not measure the e
x
act shift
area
wind spe
e
d. More
over,
usi
ng anem
o
m
eter
increases
th
e o
v
e
rall co
st, size an
d
redu
ces th
e reliabilit
y o
f
th
e sy
ste
m
. An
o
t
h
e
r
d
i
sadv
an
tag
e
of u
s
ing
an
em
o
m
eter is
that the cost and m
a
intenance
of the
anem
ometer is high. In this case, th
e
sens
orless strat
e
gies can guarantee
the
per
f
o
r
m
a
nce o
f
c
o
ntrol
s
y
stem
. In t
h
e l
iterature,
se
veral spee
d estimation tec
hni
ques are
re
ported as a
n
altern
ativ
e m
e
t
h
od
fo
r an
em
o
m
eters [6
]-[3
9]. Th
ese t
ech
n
i
q
u
e
s are b
a
sed o
n
Kalm
an
filter [9
]-[24
], n
e
u
r
al
net
w
or
k [
25]
-
[
27]
, a
d
apt
i
v
e
neu
r
o fuzzy
i
n
fere
nce sy
st
em
s [2
8]
-[
30]
, e
x
t
r
em
e l
earni
ng
m
achi
n
e [3
1]
, [32]
,
su
ppo
r
t
v
ector
m
achine [33], [34]
a
n
d etc [35]-[39].
C
once
r
ni
ng t
h
e est
i
m
a
ti
on o
f
t
h
e s
p
eed
, t
h
ere
ha
ve
bee
n
a num
b
er of resea
r
che
s
that present
e
d
di
ffe
re
nt
t
echn
i
ques
fo
r t
h
i
s
pu
rp
ose
.
Whi
l
e t
h
e est
i
m
a
tion t
e
c
hni
que
base
d o
n
o
b
se
rve
r
ap
p
r
o
x
i
m
at
el
y
depe
nds t
o
the
accuracy
of the
m
odel;
the Ka
l
m
an filter is one
of t
h
e m
o
st gene
ral m
odel
base
d estim
a
t
or [8].
The
Kalm
an filter algorithm
introduce
d
by
R
.
E
.
K
a
l
m
a
n
r
e
pr
e
s
e
n
t
s
a
n
e
f
ficient m
e
thod
for the
recursive data
pr
ocessi
ng
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
A Revi
ew
of
Wi
nd
S
p
ee
d Est
i
mat
i
o
n f
o
r Wi
n
d
T
u
r
b
i
n
e
Syst
ems B
a
se
d
o
n
K
a
l
m
an
.
...
(
M
. N
a
j
a
f
i
K
h
o
s
h
r
odi
)
1
407
The m
a
i
n
con
t
ri
but
i
o
n
of t
h
i
s
researc
h
i
s
t
o
re
vi
ew t
h
e pre
s
ent
e
d t
echni
que
s f
o
r
wi
n
d
s
p
eed
esti
m
a
t
i
o
n
b
a
sed
on
Kalm
an
filter m
e
th
o
d
.
Based
on
t
h
is,
in
th
is p
a
p
e
r, sev
e
ral
n
u
m
b
e
rs o
f
goo
d
pu
b
l
i
catio
n
s
have
re
po
rt
ed
t
h
e
wi
n
d
s
p
eed
est
i
m
a
ti
on m
e
tho
d
s a
p
pl
i
e
d
f
o
r
wi
n
d
t
u
r
b
i
n
e sy
st
em
s are di
scuss
e
d
.
T
h
e
pa
per
is o
r
g
a
n
i
zed
as fo
llo
ws:
After in
trod
u
c
ti
o
n
in
Sectio
n
1
,
th
e
Kalm
an
filter
m
e
th
o
d
o
l
o
g
y
is presen
ted
i
n
Section
2. Sect
ion
3 disc
usse
s the differe
n
t m
e
thods
of
s
p
eed estim
a
tion based
on Kal
m
an filter and
finally
t
h
e pa
pe
r c
oncl
ude
s at
Sect
i
o
n
4.
2.
KAL
M
AN FILTER
Kalm
an
filter
i
s
th
e statist
i
cal
ly seq
u
e
n
tial esti
m
a
t
i
o
n
p
r
o
c
ed
ure fo
r d
y
n
a
mic s
y
ste
m
s.
Ob
serv
atio
ns
are rec
u
rsively com
b
ined
w
ith
recen
t
forecasts with
weigh
t
s th
at
m
i
nimize the corres
p
onding
biases. This
al
go
ri
t
h
m
can be
fo
rm
ul
at
ed as f
o
l
l
o
wi
ng
e
quat
i
o
ns
[
40]
,
[
41]
.
t
t
t
t
w
x
F
x
1
.
(
1
)
t
t
t
t
v
x
H
y
.
(2)
1
1
/
.
t
t
t
t
x
F
x
(3)
t
t
t
t
t
t
W
F
P
F
P
T
1
1
/
.
.
(
4
)
1
/
1
/
.
t
t
t
t
t
t
t
t
x
H
y
K
x
x
(5)
1
1
/
1
/
.
.
.
t
T
t
t
t
t
T
t
t
t
t
V
H
P
H
H
P
K
(6)
1
/
.
t
t
t
t
t
P
H
K
I
P
(
7
)
3.
WIND SPEED ESTIMATION
B
A
SE
D ON KAL
M
AN
FILTER
In th
e literature, d
i
fferen
t
tech
n
i
q
u
e
s for
win
d
speed estimatio
n
b
a
sed
o
n
Kalm
an
filter in
cl
u
d
i
ng
lin
ear, ex
tend
ed
an
d
u
n
s
cen
t
ed
an
d etc.
Kal
m
an
filter,
h
a
v
e
b
e
en
p
r
esen
ted
.
In
Ref. [1
0
]
, a
h
ybrid
Kalm
an
filter-artificial n
e
ural n
e
t
w
ork
m
o
d
e
l was pro
p
o
s
ed
b
a
sed
o
n
an
Au
to
-Reg
ressi
v
e
In
teg
r
ated
Mov
i
ng
Av
erag
e
(AR
I
M
A
) m
o
d
e
l
t
o
fu
rt
he
r i
m
pr
o
v
e t
h
e f
o
r
e
c
a
st
i
ng acc
urac
y
of
wi
nd s
p
ee
d. T
h
e f
r
am
ewor
k o
f
t
h
e st
ud
y
used
in
[10
]
is shown
in Figure
1
.
Fi
gu
re
1.
The
f
r
am
ework
o
f
t
h
e st
udy
use
d
i
n
[1
0]
In [
1
1]
t
h
e com
p
ari
s
on bet
w
een Kal
m
an fi
lt
er and Ta
kagi
–S
uge
n
o
o
b
ser
v
er m
e
t
hods f
o
r wi
n
d
spee
d
esti
m
a
t
i
o
n
h
a
s b
een
presen
ted
.
In
th
is p
a
p
e
r, th
e Ka
lm
an
filter an
d
Tak
a
g
i
-Sug
eno
ob
serv
er tech
n
i
ques are
com
p
ared
base
d
on
re
d
u
ced
-
o
r
d
er
m
odel
s
of
a re
fe
rence
wi
n
d
t
u
r
b
i
n
e
wi
t
h
di
ffe
re
nt
m
odel
l
i
ng det
a
i
l
s
. In
[12], unscente
d Kalm
an filter
is in
corporate
d
with Support
Vector Re
gression (SVR
) based state-space
m
odel
i
n
or
der t
o
ac
curat
e
l
y
up
dat
e
t
h
e sho
r
t
-
t
e
r
m
est
i
m
a
ti
on of
wi
n
d
spee
d
seque
nce. I
n
t
h
e prese
n
t
e
d m
e
t
hod
,
support vector
regression
is used to
form
ulat
e a nonlinea
r state-
space m
o
del and the
n
unscented
Kalm
an
filte
r
is ad
op
ted
t
o
ach
iev
e
d
y
n
a
m
i
c state esti
mati
o
n
. Sch
e
m
a
t
i
c
d
i
agram
o
f
th
e u
s
ed
SVR
-
un
scen
ted
Kalm
an
filter
m
e
t
hod
f
o
r
wi
nd
sp
eed
est
i
m
a
t
i
on i
s
s
h
ow
n
i
n
Fi
g
u
r
e
2.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 6
,
N
o
. 4
,
Au
gu
st 2
016
:
14
06
–
1
411
1
408
Fig
u
re
2
.
Sch
e
matic d
i
ag
ram
o
f
th
e
SVR
-
unscen
ted Kalm
a
n
filter m
e
th
o
d
fo
r
wind
sp
eed
estim
at
io
n
[12
]
I
n
[1
3
]
, a m
e
th
od
f
o
r
w
i
nd po
w
e
r
pr
ed
ictio
n
h
a
s
b
e
en pr
opo
sed
u
s
in
g ph
ysical an
d statistical
m
odeling.
In
[13], the
Kal
m
an an
d Kol
m
ogorov–Z
urbenko filte
rs,
have
been use
d
to a
d
opt loc
a
l area
ch
aracteristics and
to
elim
i
n
ate po
ssi
b
l
e
syste
m
at
ic er
ro
rs. In
[14
]
, an
ad
ap
tiv
e Kal
m
an
filter h
a
s b
e
en
devel
ope
d a
nd applie
d to
2-metre te
m
p
erature a
n
d 10-m
e
tre wind-spee
d
forecasts in
Ic
eland.
The
pre
s
ented
m
e
thod c
o
nsists of a
d
ding two strate
gies t
o
the conve
ntional Kalm
an filte
r algorithm
that ad
aptively estim
a
te
th
e no
ise statistics in
d
i
v
i
d
u
a
ll
y.
In
[15], a c
o
mparis
on of thre
e one-ste
p
-a
he
ad fo
recasting
techniques
for
wind s
p
eed da
ta based on
robu
st Kalm
an
filterin
g
h
a
s
been
p
r
esen
ted.
Th
e au
tho
r
sho
w
ed
t
h
at th
e
weigh
t
ed
ro
bust Kalm
an
filte
r [42
]
an
d
robu
st statistics Kal
m
an
filter [43
]
yiel
d
b
e
tter
p
e
rforman
ces th
an
t
h
e stand
a
rd
Kal
m
an
filter an
d
th
e
th
resh
o
l
d
e
d Kal
m
an
filter [4
4
]
in
term
s o
f
th
e sam
p
le
sk
ewness, sam
p
le ku
rto
s
is and
residu
al error.
In [
1
7]
, t
w
o
h
y
b
ri
d m
e
t
hods
(hy
b
r
i
d
AR
I
M
A-art
i
f
i
c
i
a
l
neu
r
al
net
w
or
k
m
odel
and hy
bri
d
AR
IM
A-
Kal
m
an
m
odel) fo
r wi
n
d
spe
e
d pre
d
i
c
t
i
o
n h
a
ve bee
n
pr
o
p
o
se
d and t
h
ei
r per
f
o
r
m
a
nces have
been c
o
m
p
are
d
.
Th
e
f
r
a
m
e
w
o
rk
o
f
th
e
p
r
op
osed
h
ybr
id
m
o
dels u
s
ed
in [17
]
is show
n in
Fig
u
r
e
3
.
Fi
gu
re
3.
The
f
r
am
ework
o
f
t
h
e
pr
o
pose
d
hy
bri
d
m
odel
s
i
n
[1
7]
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
A Revi
ew
of
Wi
nd
S
p
ee
d Est
i
mat
i
o
n f
o
r Wi
n
d
T
u
r
b
i
n
e
Syst
ems B
a
se
d
o
n
K
a
l
m
an
.
...
(
M
. N
a
j
a
f
i
K
h
o
s
h
r
odi
)
1
409
Pape
r [
1
8]
p
r
e
s
ent
e
d
a t
ech
n
i
que
f
o
r t
h
e es
t
i
m
a
t
i
on
of t
h
e
wi
n
d
spee
d i
n
or
der
t
o
be
use
d
f
o
r t
h
e
co
n
t
ro
l
o
f
a
v
a
riab
le sp
eed st
all regu
lated
win
d
t
u
rb
in
e.
The alg
o
rith
m
in
[18
]
con
s
ists
of a
Kalm
an
filt
er for
est
i
m
a
ti
on of
st
at
es such as
rot
o
r spee
d and
Newt
on
-R
a
phs
o
n
m
e
t
hod
for est
i
m
at
i
on of
wi
n
d
spe
e
d (see
Fig
u
re
4
)
. In
t
h
is p
a
p
e
r, t
h
e
Kalm
an
filter is i
m
p
r
ov
ed
with
ad
ap
tiv
e al
go
rith
m
s
th
at esti
m
a
te th
e un
kn
own
cova
riances
of
the proce
ss a
n
d m
easurem
ent noises.
Fi
gu
re
4.
C
o
nt
r
o
l
sc
hem
e
used
i
n
[1
8]
In R
e
f
.
[
19]
r
o
t
a
t
i
onal
sp
ee
d re
fere
nce i
s
det
e
rm
i
n
ed b
y
aerody
nam
i
c t
o
rq
ue est
i
m
at
i
on usi
n
g
Kalm
an
filter. Ad
ap
tiv
e
o
p
t
i
m
al fu
zzy syste
m
fo
r
ro
tatio
n
a
l sp
eed
co
n
t
ro
l is presen
ted
b
a
sis of m
e
c
h
an
ical
and electrical parts of
wind t
u
rbine e
q
uations.
In [
2
1]
t
w
o m
a
ss
m
odel
base
d
est
i
m
a
t
i
on of
effect
i
v
e
wi
n
d
spee
d ha
ve bee
n
p
r
o
p
o
sed
.
I
n
t
h
i
s
pap
e
r
t
h
e gene
rat
o
r t
o
r
q
ue as sho
w
n i
n
Fi
gu
re 5 i
s
cont
r
o
l
l
e
d wi
t
h
n
onl
i
n
ea
r st
at
i
c
and dy
nam
i
c st
at
e feedbac
k
. The
esti
m
a
to
r u
s
ed
is th
e
Kalm
an
filter.
Fi
gu
re
5.
C
o
nt
r
o
l
sc
hem
e
used
i
n
[2
1]
In
[2
2
]
,
a Higher Ord
e
r Wav
e
let
Neural Network
(
HOWNN) train
e
d
with
an
ex
tend
ed
Kalm
an
filter
is proposed to
solve
the
wi
nd
forecasting
problem
. In this
pape
r, a
Kalm
a
n
filter algorithm
is used to
updat
e
t
h
e sy
na
pt
i
c
w
e
i
ght
s
of t
h
e w
a
vel
e
t
net
w
or
k
.
I
n
pa
pe
r [
2
3]
, wi
n
d
t
u
rbi
n
e s
t
at
e and
param
e
t
e
r base
d
on a
dua
l
Kalm
an
filter t
h
eory h
a
v
e
b
e
en
esti
m
a
ted
.
Th
e du
al Kalm
a
n
filter sch
e
m
e
is sh
o
w
n
in
Fi
g
u
re 6. Th
e resu
lts in
[2
3]
sh
ow t
h
at
t
h
e wi
n
d
spee
d an
d r
o
t
o
r spe
e
d i
n
bel
o
w a
n
d ab
ov
e rat
e
d
wi
n
d
spee
d ca
n be est
i
m
at
ed
wi
t
h
a
h
i
gh
q
u
a
lity.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 6
,
N
o
. 4
,
Au
gu
st 2
016
:
14
06
–
1
411
1
410
Fig
u
re
6
.
Du
al
Kalm
an
filter sch
e
m
e
[2
3
]
4.
CO
NCL
USI
O
N
To
con
t
ro
l th
e
v
a
riab
le sp
eed
wind
t
u
rb
in
es,
th
e
wind s
p
ee
d m
easurem
ent
is re
quire
d.
Norm
ally, the
wind s
p
eed provide
d
by the anem
o
m
eter. Usi
n
g anem
om
eter increases
the syste
m
cost,
m
a
intenance
,
co
m
p
lex
ity
an
d
redu
ces th
e
reliab
ility. Accu
rate
wind
sp
eed
estim
a
tio
n
fo
und
to
b
e
u
s
efu
l
in
term
s o
f
g
r
o
w
t
h
in ene
r
gy capt
u
re
. In t
h
is
re
search, a
re
view
has
be
e
n
do
ne
o
n
t
h
e e
ffect
i
v
e
est
i
m
a
t
i
on
of
wi
nd
spee
d.
Practically all
th
e i
m
p
o
r
tan
t
wind
sp
eed
esti
m
a
t
i
o
n
tech
n
i
q
u
e
s b
a
sed o
n
Kalm
an
filter ap
p
lied
t
o
wi
n
d
turbines ha
ve been
ca
refully
discusse
d.
REFERE
NC
ES
[1]
G.
M.
Jose
lin He
rbe
r
t
, et a
l
.
, “A review of wind
energ
y
technolo
g
ies,”
Renewab
l
e and sustainable energy
Review
s
,
vol. 11
, pp
. 1117
-1145, 2007
.
[2]
A. K. Sharma, “
S
tud
y
of wind
turbine based
seig
under balanced/unbalanced
lo
ads and ex
citation,”
Int
e
rnation
a
l
Journal of Electrical and
Computer
Eng
i
neer
ing
,
vol. 2
,
pp
. 353-3
70, 2012
.
[3]
Y.
Zheng
, et al.
, “Mode
Analy
s
is
of Horizontal Axis Wind Turbine Blades,”
TELKOMNIKA Indonesian Journal of
Ele
c
trica
l
Eng
i
n
eering
, vol. 12
,
pp. 1212-1216
,
2014.
[4]
H.
Li
, et al.
, “Fault-to
ler
a
nt con
t
rol for curren
t
sensors of
doubly
fed indu
ction generators based
on an improved
fault detection m
e
thod,”
M
e
asurement
, vo
l. 47, pp
. 929-937
, 2014
.
[5]
E. Sesto
and C.
Casale, “E
xploitation of
wind as
an en
erg
y
source
to m
e
e
t
th
e wo
rld's
ele
c
tr
ici
t
y
d
e
m
a
nd,”
Journal
of Wind
Engin
e
ering and Industr
ial
Aerodynamics
, vol. 74, pp. 37
5-387, 1998
.
[6]
M. N. Soltani
, et al.
, “
E
s
tim
atio
n of rotor eff
ect
ive wind s
p
eed
:
A com
p
aris
on,”
IEEE Transactions on Contro
l
Systems Technology
, vo
l. 21, pp.
1155-1167, 201
3.
[7]
K.
Z.
Østergaard
, et al.
, “
E
s
tim
ation of eff
ect
iv
e wind s
p
eed,”
I
n
Journal of Ph
ysics: Conference Series
, vol. 75,
pp. 012082
, 200
7.
[8]
T. Bur
t
on
, et al.
, “Energ
y
Handb
ook,”
John W
i
ley
&
Son
s
, 2011.
[9]
H. Baba
zadeh
, et
al.
, “An hour ahead wind speed
prediction b
y
K
a
lman filter
,”
In
Power Electronics and Machin
es
in Wind
Applica
t
ions (
PEMWA)
,
pp. 1-6
,
2012
.
[10]
O. B. Shukur and M. H.
Lee, “Daily
wind speed forecasting thr
ough h
y
brid KF-ANN
model based on ARIMA,
”
Renewab
l
e Ener
gy
, vo
l. 76, pp. 6
37-647, 2015
.
[11]
E. Gauter
in
, et al.
, “
E
ffe
ctiv
e
wind speed esti
m
a
tion: Com
p
aris
on between Kalm
an Filter an
d Takagi–Sugen
o
obs
erver t
echn
i
q
u
es
,”
ISA transactions
, 2015.
[12]
K. Chen and J. Yu, “Short-term wind speed prediction us
ing an
unscented Kalm
an filt
er based state-space suppor
t
vector
regr
ession approach,”
Ap
plied
Energy
, vo
l. 113
, pp
. 690-7
05, 2014
.
[13]
C. Stathopou
los
, et al.
, “Wind
power predictio
n based on nu
me
rical
and st
at
istica
l
m
odels,
”
Journal of Wind
Engineering and
Industrial
Aero
dynamics
, v
o
l. 1
12, pp
. 25-38
, 2
013.
[14]
P. Crochet
,
“
A
daptiv
e Kalm
an filter
i
ng of 2-m
e
tre tem
p
er
ature
and 10-m
e
tre wind-speed fore
ca
sts in Iceland
,
”
Meteorolog
ical Applica
tions
, vo
l. 11
, pp
. 173-18
7, 2004
.
[15]
C. D. Zuluaga
, et al.
, “Short-term
wind
speed prediction based on
robust
Kalm
an filtering
: An experim
e
ntal
com
p
aris
on,”
Ap
plied
Energy
, vo
l. 156
, pp
. 321-3
30, 2015
.
[16]
F. Cassola
and
M. Burlando
, “
W
ind speed an
d wind en
erg
y
forecast
through
Kalm
an fi
lter
i
ng of Num
e
ric
a
l
W
eather
P
r
edic
ti
on model outpu
t,”
App
lied
en
er
gy
, vol. 99
, pp
. 154
-166, 2012
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
A Revi
ew
of
Wi
nd
S
p
ee
d Est
i
mat
i
o
n f
o
r Wi
n
d
T
u
r
b
i
n
e
Syst
ems B
a
se
d
o
n
K
a
l
m
an
.
...
(
M
. N
a
j
a
f
i
K
h
o
s
h
r
odi
)
1
411
[17]
H.
Liu
, et
a
l
.
,
“Comparison of two new ARI
M
A-ANN and
AR
IMA-Kalma
n
h
y
br
id metho
d
s for wind speed
prediction,”
App
lied
Energy
, vo
l. 98, pp. 415-424
, 2012
.
[18]
D. Bourlis and J. A. M. Blei
js, “
A
wind speed estim
ati
on m
e
tho
d
using adaptiv
e
Kalm
an filter
i
n
g
for a variab
le
s
p
eed s
t
all r
e
gul
ated wind turbin
e,”
In 11th International Confer
ence on
Probabilistic M
e
thods Applied to Pow
e
r
Sy
ste
m
s (PMAPS)
, pp. 89-94
, 20
10.
[19]
Z. Xin-fang
, et al.
, “Adaptive o
p
timal fuzzy
co
ntrol for vari
able speed fixed pitch wind turbines,”
In Fifth World
Congress on Int
e
llig
ent
Control
and Automat
i
on
(
W
CICA
)
, pp. 24
81-2485, 2004
.
[20]
H. Vihriälä, “Co
n
trol of
var
i
able
s
p
eed wind
turbi
n
es
(P
hD thes
is
),
”
Tampere University o
f
techno
logy
, 2002
.
[21]
B. Boukhe
zzar
and H. Siguerd
i
d
jane
, “
N
online
a
r cont
rol
of
a
variab
le-spe
ed
wind turbine
using a
two-m
a
ss
model,”
IEEE Transactions on
Energy Conversion
, vol. 26
, pp
. 14
9-162, 2011
.
[22]
L. J
.
Ri
cald
e
, et al.
, “Higher
or
der wavelet neu
r
al n
e
tworks with
Kalman learn
i
ng for wind speed forecasting,”
In Symposium o
n
Computationa
l Intelligen
ce
Ap
plications In Sm
art Grid (
C
IASG)
, pp. 1-6
,
2011
.
[23]
P. Mate
ljak
, et
al.
, “Dual kalman estimation
of wind
tu
rbin
e s
t
a
t
es
and
pa
ram
e
ters
,
”
In
Pr
oceed
ings
of
th
e
International Co
nference on
Pro
cess Control
, pp. 85-91, 2011.
[24]
D. Jena and S. Rajendr
an, “A review
of estimation of effective wind speed
based control of wind turbines,”
Renewab
l
e and
Sustainable Energy Reviews
, vol. 43
, pp
. 1046-1
062, 2015
.
[25]
H.
Li
, et
al.
, “Neural-network
-
based
sensor
less
maximum
wind energ
y
capture with
co
mpensated power
coeffi
cien
t,
”
I
E
EE Transactions
on Industry App
l
ications
, vol. 41
, pp. 1548-1556,
2005.
[26]
O. Barambones
, et al.
, “A neural network based
wind speed estimator for a wind turbine con
t
ro
l,”
In
15th
IEEE
Mediterranean Electrotechnica
l
Confer
en
ce
(
M
EL
ECON)
, pp. 1
383-1388, 2010
.
[27]
F. Jaram
illo-Lop
ez
, et
al.
, “A novel onlin
e training neural network-ba
sed algorithm for wind
speed estimation
an
d
adapt
i
ve con
t
rol
of P
M
S
G
wind turbine s
y
s
t
em
for m
a
xim
u
m
power extra
c
tion
,
”
Ren
e
wable En
ergy
,
vo
l. 86
, pp
.
38-48, 2016
.
[28]
D. Petkovi
ć
, et
al.
, “Gener
alized adaptiv
e neuro
-fuzzy
based me
thod for wind speed distribution
prediction,”
Fl
ow
Measurement an
d Instrumentatio
n
, vol. 43
, pp
. 47
-52, 2015
.
[29]
S.
Sha
m
shirba
nd
, e
t
al
.
, “Sensorless estimation of
wind speed
b
y
adaptiv
e neuro-fu
zzy
methodo
log
y
,”
In
ternationa
l
Journal of Electrical
Power
&
E
n
ergy Systems
, v
o
l. 62
, pp
. 490-4
95, 2014
.
[30]
E. T. Al-Sham
m
ari
, et al.
, “
E
stim
ation of wi
nd turbin
e wake effect b
y
adaptive neuro-
fuzzy approach,”
Flo
w
Measurement an
d Instrumentatio
n
, vol. 45
, pp
. 1-
6, 2015
.
[31]
S.
Wu
, et al.
, “
E
xtrem
e
l
earnin
g
m
achine bas
e
d wind s
p
eed es
timation and sensorless control for wind turbin
e
power gen
e
ratio
n
s
y
stem,”
N
e
urocomputing
, vol. 102, pp. 163-17
5, 2013
.
[32]
G.
B.
Huang
, et al.
, “Extreme learning machine:
theor
y
and app
l
ications,”
Neuro
c
omputing
,
vol. 70, pp. 489-501,
2006.
[33]
X.
Kong
, et al.
, “Wind speed prediction using reduced sup
port vector machines with feature selection
,
”
Neurocomputing
, vol. 169
, pp
. 44
9-456, 2015
.
[34]
D.
Liu
, et al.
, “Short-term wind s
p
eed for
e
castin
g
using wavelet tr
ansform and s
upport vector machines optimized
b
y
genetic algorithm,”
R
e
newab
l
e En
er
gy
, vol. 6
2
, pp
. 592-597
,
2014.
[35]
W.
Qi
a
o
.
, “Echo
-
state-n
e
twork-b
a
sed real-time w
i
nd
speed estimation for wind power generation
,
”
In International
Joint Conferen
ce on
Neural Networks (
I
JCNN)
, pp. 2572-2579
,
2009.
[36]
H.
Liu
, et al.
, “An EMD-recursiv
e ARIMA method to pred
ict win
d
speed for
railway
strong wind
warning s
y
stem,”
Journal of Wind
Engineeri
ng and
Industrial
Aero
dynamics
, v
o
l. 1
41, pp
. 27-38
, 2
015.
[37]
L. T
i
an
, et al.
,
“A Gaussian RBF network based wind speed
estimation
algor
ithm for maximum power point
track
ing,”
En
ergy Proc
edia
, vol.
12, pp
. 828-836
, 2011.
[38]
E.
Erdem and
J. Shi, “ARMA based appro
ach
es for
forecasting
the tup
l
e of win
d
speed
and dir
ection,”
App
lie
d
Energy
, vol. 88
,
pp. 1405-1414
,
2011.
[39]
R. G. K
a
vas
s
e
ri
and K.
S
e
e
t
har
a
m
a
n, “
D
a
y
-ah
e
ad wind s
p
e
e
d f
o
recas
t
i
ng us
ing
f-ARIM
A
m
o
dels
,”
Renewable
Energy
, vol. 34
,
pp. 1388-1393
,
2009.
[40]
R. E.
Kalm
an,
“
A
new approach
to lin
ear f
ilterin
g and pred
iction
problems,”
Jour
nal of basic Eng
i
neering
, vol.
82
,
pp. 35-45
, 1960
.
[41]
P. Louka
, et
al.
, “
I
m
p
rovem
e
nts
in wind s
p
eed forecas
ts
for
wind power prediction purpos
e
s
us
ing Kalm
an
filte
ring,
”
Journal of Wind
Engin
eering and
Indus
trial Aerodynam
ics
, vo
l. 96, pp.
2348-2362, 200
8.
[42]
J.
A.
Ting
, et al.
,
“
A
Kalm
an filt
er for robu
st outli
er de
tec
t
ion,”
In IEEE/RSJ
Internati
onal Conference on
Intelligen
t Robo
ts and Systems
, p
p
. 1514-1519
, 2
007.
[43]
T
.
Ci
pra a
n
d R.
Rome
ra
,
“Ka
l
man fi
lt
e
r
with ou
tliers and
missing observations,”
Te
s
t
, vo
l. 6, pp. 3
79-395, 1997
.
[44]
I. C. Schick and
S. K. Mitter, “Robust recursive
estim
a
tion in th
e presence of heav
y
-
t
a
il
ed observ
a
tion noise,”
The
Annals of Statistics
, pp
. 1045-10
80, 1994
.
Evaluation Warning : The document was created with Spire.PDF for Python.