TELKOM
NIKA
, Vol. 13, No. 4, Dece
mb
er 201
5, pp. 1170
~1
178
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v13i4.2049
1170
Re
cei
v
ed Ma
y 21, 201
5; Revi
sed O
c
tob
e
r 1, 2015; A
c
cepted O
c
to
ber 15, 20
15
Dynamic Stability Improvement of Multimachine Power
Systems using ANFIS-based Power System Stabilizer
Agung Budi Muljono*
1
, I
Made Ginarsa
2
, I Made Ari Nrartha
3
1,2,
3
Dept. of El
ectrical En
gi
neering, Mataram
Universit
y
Jln. Maja
pah
it No. 62 Matara
m, Indonesi
a
T
e
lp/fa
x +
62 3
7
0
636
75
5
e-mail: a
gun
gb
m@unram.ac.i
d
1
; kadekgi
n@
ya
ho
o.com
2
; ari.nrartha
@
gma
il.com
3
A
b
st
r
a
ct
Moder
n p
o
w
e
r
system
are
ver
y
vurn
erab
le
to
ag
ains
t
lo
ad
fl
uctuatio
n
duri
n
g the
i
r
oper
atio
n. Lo
ad
fluctuatio
n is i
dentifi
ed as s
m
a
ll di
st
urba
n
c
e that it is very importa
nt
in sma
ll sig
n
a
l
stability (dyn
a
m
ic
stability)
testing. This r
e
search
consisted of large scale pow
e
r system
s
y
m
p
lificati
on.
And, ANFIS-bas
e
d
pow
er system
stabili
z
e
r (AP) is pr
oposed to im
pr
ove the
dynam
i
c stabilit
y of m
u
lti
m
ac
hine. The ANFIS
meth
od is pr
op
osed b
e
ca
use
the ANF
I
S comp
utatio
n is
more efective th
an Ma
md
ani fu
zz
y
co
mp
utatio
n
.
Simulation res
u
lts show t
hat the proposed PSS is abl
e t
o
m
a
intain the dynam
i
c stability by decreas
ing
peak ov
ersho
o
t
to the value
3,37
10
5
pu a
nd acce
lerati
ng
settling time to the time 4.01
s for rotor spee
d
devi
a
tion of M
a
chi
ne-2. Also,
the peak
over
shoot is decr
e
ased to the val
ue
1,3
4
10
5
pu an
d the settlin
g
time is acc
e
l
e
r
a
ted to the time 3.98 s for rotor spee
d dev
ia
tion of Machi
n
e
-
3.
Ke
y
w
ords
: Stability improv
em
ent, dynam
i
c,
m
u
ltimachine,
PSS, ANFIS
Copy
right
©
2015 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Mode
rn p
o
wer sy
stem
s a
r
e cover l
a
rg
e are
a
with
some ge
ne
rati
on unit
s
con
necte
d to
bulk
pow
e
r
sys
tems
via trans
miss
ion sys
tem. Meanw
hile, al
mo
st of co
nsume
r
s a
r
e l
o
cated
in
city su
ch
as: come
rcial
and offici
al
compl
e
x
are
a
s,
subu
rb
a
n
area, ind
u
s
trial
are
a
a
nd
resi
den
sial a
r
ea. Analysi
s
of a large power
sy
ste
m
is very complicated a
nd difficult. So,
simplifiying
schem
e
sho
u
l
d
be
do
ne to
solve thi
s
di
fficulties probl
em. This large
system
is divi
de
into so
me
small op
eratio
n area
s [1]. The b
a
lan
c
in
g of ele
c
tri
c
al-me
c
h
ani
ca
l ene
rgy is very
important to keep the machine
works in
synchronous
mode. Rotor oscillation is a problem
durin
g p
o
wer system
op
eration that a
p
peared
of th
e
roto
r o
s
cillation d
ue to l
o
a
d
fluctu
ation
at
load buses.
Power sy
stem stabili
zer (PSS) is appli
ed to dam
p
the rotor oscil
l
ation problem.
Rob
u
st
H-i
n
finity loop
sha
pping te
ch
niq
ue is pr
opo
se
d to da
mp a
singl
e ma
chi
ne in l
a
rg
e
scale
power s
y
s
t
ems
.
The proposed
PSS ensures
to
cover a
s
e
t of pertub
operating points
with
respect to the nominal sy
stem and abl
e to ma
intain the whol
e system stabilit
y [2]. Optimized
para
m
eter P
SS is u
s
e
d
to
mitigate th
e
synchro
nou
s
gene
rato
r o
scillation. Where, a p
e
rfo
r
ma
nce
index is ta
ken autom
atically by mo
nitori
ng th
e
gene
rator
para
m
eter.
Furthe
rmo
r
e,
the
asse
ssm
ent of
the perfo
rmance
i
ndex is
o
b
taine
d
u
s
ing
an
expe
rt sy
stem
ba
sed o
n
wavefo
rm
record of gen
erato
r
param
eter [3].
Identifier
a
n
d
co
ntrolle
r scheme ba
sed
on
a
r
tifi
cial i
n
telligent al
gorithm su
ch
a
s
:
Ne
ural
netw
o
r
k
,
fuzzy logic and neur
o-
fuz
z
y
c
o
ntr
o
ller
s
applied to
elec
tr
ical
and other engineer
ing fields
are very popular in
recent
years. A neuro id
entifier-model reference a
daptive
controller PSS is
applie
d in sin
g
le machine
and multima
c
hine with onli
ne adju
s
ted [4]. An ANFIS model is abl
e to
estimate the online critical clearing ti
me (CCT
) of transient stability in multimachine power
system.
Whe
r
e, the ANFIS
model
gives
the CCT q
u
ite sati
sfied
with high
accu
ra
te solutio
n
an
d
low
com
putat
ion time [5].
Suppo
rt vect
or m
a
chine
(SVM) metho
d
is
appli
ed t
o
tran
sie
n
t st
ability
clasification by
Maulin
et
al
. It is o
b
tained t
hat th
e SVM meth
od give
s b
e
tter result tha
n
multilayer perceptron
-
ne
ural
net
wo
rk (M
LP-NN) meth
od [6]. PID-SVC b
a
sed
on
recurrent
ne
u
r
a
l
netwo
rk
(RNN) ha
s
b
een applie
d
to co
ntrol cha
o
s
a
nd voltage
co
llapse in
a p
o
we
r
system
[7].
Also, ANFIS-based compo
s
ite co
ntrolle
r-SVC an
d PID-lo
op have
been a
pplie
d to control ch
a
o
s
and volta
ge
colla
pse a
n
d
to regul
ate
the voltage
at load
bu
s
with lo
adin
g
fluctuation
[8
][9].
ANFIS controller i
s
u
s
ed
to mai
n
tain
dynamic
resp
onse of
HV
DC
system
[1
0]. Furthe
rm
ore,
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 117
0 – 1178
1171
ANFIS power sy
stem
stabilizer
(PSS) has been applied to im
pr
ove the st
ability of single
machi
ne based on feedback linea
rization [11]. And, adaptive fuzzy rul
e
-based PSS [12] a
n
d
fuzzy l
ogi
c P
SS [13] [14]
are al
so
used to ma
i
n
tain dynamic stability
of
a power
system.
Some
probl
em a
c
co
unts in the la
rge scale
po
wer sy
stem
a
r
e
method to si
mplify the large scale
po
wer
system and control schem
e
to
improve
stability of the system.
In orde
r to simplify compl
e
x powe
r
system, this system wa
s bro
k
en down into 3 (thre
e
)
areas
(Area I,
Area II and
Area III). Then, the mu
ltimachi
ne in Area I
was regul
ated by apply
i
ng
ANFIS-based PSS to improve its dy
namic stability.
This paper
i
s
organized as
follows: Dynamic
stability of a multimaci
ne
power sy
stem is descr
ibed in Section
2. C
onventional po
wer system
stabili
zer de
sign a
nd A
N
F
I
S algorithm
are
detaile
d
in Sectio
n 3
and
4,
re
sp
ectively. Nex
t
,
simulatio
n
re
sult an
d a
nal
ysis a
r
e
presented in
Se
cti
on 5. And, th
e co
ncl
u
si
on i
s
p
r
ovide
d
in
the
last
se
ct
ion.
2. D
y
namic
Stabilit
y
of a Multimachine Po
w
e
r S
y
s
t
em
A multimachi
ne po
we
r sy
stem in this
resea
r
ch
is g
i
ven by Padiyar [15]. This system
con
s
i
s
t of 3
9
-bus,
10
-ma
c
h
i
ne, an
d thi
s
system
is sho
w
n i
n
Fi
gure
1. The
sy
ste
m
was sepa
rated
into Area I wit
h
Mac
h
ine-1,
Mac
h
ine-2 and Mac
h
ine-3.
Area II:
Mac
h
ine-4, Mac
h
ine-5,
Machine-
6 and Machine-7. And, Area III: Mach
ine-8, Machine-9
and Machine-10.
Stability of the multimachi
ne in Area I
is fo
cused
on this
research included electro-
mech
ani
cal i
n
tera
ction in
3-ma
chi
ne m
odel. The
r
ef
o
r
e, the Ma
chi
ne-1
at Bus
1
wa
s thre
ated
as
a refe
ren
c
e/swing b
u
s. Fu
rthermo
re, the
spee
d an
d a
ngle rotor d
e
v
iation of the Machi
n
e
-
1 was
taken a
s
ze
ro, resp
ectivel
y
. Mechani
ca
l and rea
c
tive mode eq
ui
pped by exci
ter tipe IEEE 1a
were used to rep
r
e
s
ent the
classi
cal mo
del of each machi
ne.
Figure 1. Single line diag
ram of a multimachi
ne po
wer sy
stem
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Dynam
ic Stability Im
provem
ent of Mu
lti
m
achine Power System
s using …
(
A
g
ung
Bu
d
i
Mu
ljo
no
)
1172
Figure 2. Linear model of ith
mac
h
ine equipped by ANFIS-PSS in
a
multimachine power s
y
s
t
em
Stability is the ability of
power sy
stem t
o
co
ver the disturbance
at
norm
a
l operation the
effort to maintain the power syste
m
go
ing to
steady
state after the di
stu
r
ba
nce is diapp
eared.
Small sign
al (dynami
c
)
stability includ
ed one o
r
so
me machine
s
were
cha
n
g
ed the ope
ra
ting
point mod
e
ra
tely. Dynamical b
ehavio
r of the sy
ste
m
is de
pen
d
ed on i
n
tera
ction of tu
rbi
ne,
gene
rato
r, also the
cont
rol
l
er characte
ri
stic
su
ch a
s
govern
o
r a
n
d
excitation sy
stem
s. Form
ulas
for rep
r
e
s
ent
ed the dynam
ical sy
stem of
the ith
machi
ne in linea
r m
odel are as fo
llows [16]:
∆
∆
,
(
1
)
∆
∆
∆
∆
,
(
2
)
whe
r
e
∆
,
∆
,
,
,
∆
and
∆
are the mechani
cal torque, el
ectri
c
al torqu
e
, inertia
con
s
tant, da
mping
con
s
ta
nt, rotor spee
d and
rotor
a
ngle deviatio
n
of maci
ne i
t
h, resp
ective
ly.
And
is the synch
ron
o
u
s
sp
eed. Synchro
nou
s ma
chi
n
e compo
nent
s to repre
s
e
n
t
the dynamic
stability an
alysis a
r
e
divide
into m
e
chan
ical
and
rea
c
tive com
pone
nt (mo
d
e
)
. T
he m
e
chani
cal
and
reactive
mode to ill
ust
r
ate a
multim
achi
ne power system
equi
pped
by PSS in linear m
o
del
is illustrat
ed
by diagram
block in
Figure 2. This linear system
can
be describ
ed
by state space
or
Lapla
c
e form for time or fre
quen
cy doma
i
n, resp
ectivel
y
. The model formula
s
are as follo
ws:
State spa
c
e:
∆
∆
∆
∆
∆
∆
; Laplace:
∆
∆
0
∆
∆
∆
∆
∆
(3)
3. Conv
entional PSS
The functio
n
of the PSS is to provide dam
pin
g
torqu
e
com
p
o
nent to the gene
rato
r
(ma
c
hin
e
) rotor o
scillatio
n
by regulatin
g
its ex
citation
system thro
ugh an a
dditi
onal sta
b
ilizi
n
g
signal. To provide the damping
torque, the stabilizer must prod
uce a
com
ponent of elect
r
i
c
al
torque in phas
e
with the rotor
s
p
eed deviati
on. The PSS devic
e
is
very important to improve
stability of ov
erall
power systems. Since the porp
use of a PSS i
s
to introduce
a dam
ping torque
comp
one
nt, a logical si
gn
al to use fo
r
regul
ating ex
citation
syste
m
of machin
e is rotor
sp
eed
deviation. And, PSS output is an
additional
stabili
zi
ng si
gnal
(
V
s
). Conventional PSS device
con
s
i
s
t of g
a
in, wa
sh
out
and p
h
a
s
e
comp
en
satio
n
blo
c
ks. Th
e gain
blo
c
k determi
ne
s
the
amount
of d
a
m
ping i
n
tro
d
u
c
e
by the PS
S. The
sign
al
wa
sh
out bl
o
c
k serve
s
as
a hig
h
fre
que
ncy
filter, with the time c
o
ns
tant
T
w
. The pha
se
comp
en
sa
tion bo
ck
pro
v
ides the
app
rop
r
iate p
h
a
s
e
lead characte
ristic to
comp
ensat
e the p
hase lag bet
wee
n
exciter
input and g
e
nerato
r
(air-g
ap)
electri
c
al torque. Diagra
m
block of conventional PSS
is shown in Figure 3(a).
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 117
0 – 1178
1173
4. Adap
tiv
e
Neuro F
u
zz
y
Inferenc
e Sy
stem
Adaptive n
e
u
r
o-fu
zzy infe
rence
system
is o
n
e
of m
e
thod
ba
sed
o
n
a
r
tificial i
n
telligent.
The ANFIS method fun
c
tion is sa
me a
s
the fu
zzy ru
le base
d
on Sugeno al
gorithm. The ANFIS
is con
s
i
s
t
of premi
s
and
consequ
en
ce para
m
eter
s.
So, both the
para
m
eters
were o
b
taine
d
by
off-line l
earni
ng p
r
o
c
e
s
se
s with
lea
s
t
sq
uare
s
e
s
timation
(LSE) an
d ba
ckp
r
opa
g
a
tion al
gorith
m
s.
At forward st
ep, the pa
ra
meters were
identifi
ed by
usin
g LSE m
e
thod. Mo
reo
v
er, at backward
step, the error sig
nal wa
s attenuated
back,
an
d the paramete
r
s were mai
n
tained by u
s
in
g
gradi
ent de
scent optimizati
on.
Suppo
se
that the ANFIS network ha
s 2 (two) inp
u
ts
x
,
y
a
nd
an
output
O
. T
h
i
s
ANFIS mod
e
l ha
s 2
rule
s a
nd b
a
sed
on first-o
r
de
r
fuzzy Su
gen
o
.
The
rule
s a
r
e
as
follows
[17]:
Figure 3. PSS bloc
k
diagram
1: If
is
and
is
Then
2: If
is
and
is
Then
,
Finally, outpu
t the ANFIS network is follo
w:
∑
∑
∑
,
After the ANFIS algorith
m
has be
en
built. Ther
efo
r
e, the ANFIS-ba
sed PS
S in this sch
eme
control was
applied to replace the function
of the
conventional
PSS. The ANFIS-based
PSS
block diagram is illust
rated in Figure 3(b).
5. Building Process of ANFIS-
based
Po
w
e
r S
y
stem Stabiliz
er
Before the A
N
FIS-bas
e
d
PSS is
applied to
a multimac
hine
s
y
s
t
em
, the proposed PSS is
desi
gne
d and
con
s
tru
c
ted
by some lea
r
ning proce
sses in off-line
mode. Data t
r
ainin
g
that used
for thi
s
le
arning
pro
c
e
ss we
re
obtai
ned
by si
m
u
lating th
e
multimachine
equi
ppe
d
with
conventional
PSS. To obtain the data training, a
m
u
ltimachi
ne syst
em
with conv
entional
PSS is
force
d
by
sin
g
le an
d multi
p
le ste
p
fun
c
t
i
ons.
Wh
ere
the
step fun
c
ti
on was u
s
ed
to impleme
n
t
the
cha
nge of m
e
ch
ani
cal torque in the m
a
chi
ne due to
load fluctuati
on. In this learning p
r
o
c
e
s
s, a
4000-data training
set
was used to desi
gn the
ANFIS-based PSS. The i
nput
s
of ANFIS-based
PSS were rot
o
r speed devi
a
tion (
)
and its derivative (
∆
). And, the output was a
n
additional
stabili
zing signal
(
V
s
).
Structure of the ANFIS PSS model
was built by using 7 (seven)
Gau
ssi
an me
mbershi
p
fun
c
tion
s for the
input and 4
9
(forty-nin
e) rules fu
zzy Su
geno o
r
d
e
1 for
the output, re
spe
c
tively. After some le
arning pr
ocesses we
re cond
ucted, the Su
geno fuzzy form
of the PSS w
a
s built automatically. Thi
s
Sugeno fu
zzy form of the PSS is illustrated in Fi
gure
3(c). And, control surface of
respective
input-output the
ANFIS-based PSS was obtained. A set
of input-o
utp
u
t control su
rface
wa
s o
b
t
ained a
s
foll
ow:
-
∆
-
V
s
. This in
put-o
utput co
ntrol
surfa
c
e of PS
S is sho
w
n in
Figure 3
(
d
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Dynam
ic Stability Im
provem
ent of Mu
lti
m
achine Power System
s using …
(
A
g
ung
Bu
d
i
Mu
ljo
no
)
1174
6. Results a
nd Analy
s
is
To dem
on
strate the pe
rfo
r
man
c
e
of a
mult
imachine
power
syste
m
, this multi
m
achi
ne
system
wa
s
examined
u
s
ing Matla
b
/Simulink 7.
9.0.
529
(R2009
b
)
[18] on
an I
n
tel Co
re
2
Duo
E6550 23
3 GHz PC
comp
uter and wi
nd
ows 7 64-bi
t
(win
64
) ope
rating system.
The simulati
ons
were don
e as follows:
Figure 4. Improveme
n
t of the roto
r sp
ee
d deviation o
n
a singl
e dist
ruba
nce
6.1. Perform
a
nce of Proposed PSS at a Single Disturbance
A multimachi
ne po
we
r sy
stem Area I
is ru
n with
ou
t any control
scheme.
Ne
xt, the
system i
s
equipped by
2 (two)
conventional PSS(s)
at Mach
i
ne-2 and Machine-3. And, 2 (t
wo)
ANFIS-based (proposed)
PSS(s)
were applied at
re
spective m
a
chine to mai
n
tain the
system
respon
se
s. F
i
rst S
c
en
ario
, the syste
m
wa
s fo
rced
by a
singl
e
distu
r
ba
nce
at ma
chin
e
-
2
(me
c
ha
nical torqu
e
d
e
viation,
T
1
)
at th
e value
of 0.1
pu a
nd th
e ti
me of 0.1
s. T
he respon
se
s of
the sy
stem
were
ob
se
rved
at thei
r
rotor
spe
ed
and
an
gle d
e
viation.
The
s
e
si
mul
a
tion
re
sults
are
illustrate
d in Figures 4, 5, and liste
d in Table 1.
Figure 4(a) and Table 1
show the propos
ed PSS (AP) was
able
to improve the peak
overshoot (P
o) of
the roto
r
sp
eed
deviat
i
on (
2
) at the value
of
3.
37
10
5
pu. T
he
settling ti
me
(St) was al
so improved
at
the time of 4.01 s. Meanwhile
, the conv
entional PSS (CP) and the
multimachine without any
PSS (WP
)
gave th
e peak
overs
h
oot at the values
of
4.
65 and
5.50
10
5
p
u
, re
spe
c
tivel
y
. So, the se
ttling time of
the CP
and
WP were a
c
hived at time
s of
6.45 a
nd
>2
0 s. Fi
gu
re 4
(
b) an
d Ta
bl
e 1 ilu
stra
te
the pe
ak
ove
r
sh
oot im
pro
v
ement of
ro
tor
spe
ed deviati
on (
3
) by
the pro
p
o
s
ed P
SS at the value of
1.34
10
5
pu. Also, the settling ti
me
of the
3
was
ac
hieved at the value of 3.98 s
fo
r t
he propos
ed
PSS. While, when the s
ystem
wa
s eq
uipp
e
d
by the
CP
and
WP, the
system
a
c
hi
eved the
pea
k ove
r
shoot
at the value
s
of
1.83
an
d
2.
21
10
5
pu, res
p
ec
tively. And,
the other
PSS(s
)
ac
hieved the
s
e
ttling time at times
of 6.39 and >20 s.
The pea
k ov
ershoot of rot
o
r angl
e devi
a
tion (
2
) was al
so maint
a
ined by the
prop
osed
PSS at the value of
0.46
.
The
CP a
nd
WP gave
the
pea
k
ove
r
sho
o
t at the valu
es
of
0
.
56
an
d
0.67
, re
spe
c
tively. More
over, the ste
ady state of
the roto
r ang
le deviation
wa
s a
c
hieve
d
at
0.34
. The
s
e
ttling time
of the proposed PSS was
at
time of 4.32 s
.
Meanwhile, the settling
time of the other PSS(s) were at times
of 5.
37 and >20
s, respectively
. These responses
are
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 117
0 – 1178
1175
illustrated in
Figure 5(a) and T
able 1. Fi
gure 5(b)
and Table 1 show the responses of
rotor angle
deviation for
Machi
n
e
-
3 (
3
). It is described that the pea
k ov
ersho
o
t was a
c
hi
eved at the valu
e
of
0.13
5
. T
he pe
ak
oversho
o
t for the
CP
and
WP
were at the v
a
lue
s
of
0.1
63 an
d
0.19
5
.
The roto
r ang
le steady stat
e (Ss)
was a
c
hived at the value of
0.09
4
for all PSS
(s). The s
e
ttling
time of the propos
ed PSS was
achiv
ed
at time of 4.09 s
.
Meanwhile,
the s
e
ttling time of the CP
and WP
were
obtained at time of 6.32 an
d >20
s.
Figure 5. Improveme
n
t of the roto
r
angl
e deviation o
n
a singl
e dist
urba
nce
Table 1. Responses
of a system without
(WP), with
conv
entional (CP) and
ANF
I
S (AP) PSS(s).
PSS
2
2
Peak overshoot (
P
o)
10
5
(pu
)
Settling time
(St) (s
)
Po
(
)
St
(s
)
Ss
(
)
WP
5.50
> 20
0.67
> 20
0.34
CP
4.65
6.45
0.56
5.37
AP
3.37
4.01
0.46
4.32
PSS
3
3
Po
10
5
(pu
)
St
(s
)
Po
(
)
St
(s
)
Ss
(
)
WP
2.21
> 20
0.195
> 20
0.094
CP
1.83
6.39
0.163
6.32
AP
1.34
3.98
0.135
4.09
In Firs
t Sc
enario, it is
s
hown that the propos
e
d PSS is
able to giv
e
better performanc
e
than the
other PSS. Where, the
proposed PSS produces peak ov
er
shoot val
u
es of
rotor speed
and an
gle a
r
e less than t
hat the pea
k overshoot of
the other P
SS. Also, settling time of the
proposed PSS of all responses
are shorter than that
the settling ti
me of the others.
6.2. Perform
a
nce of Proposed
PSS at Multiple Disturbances
Secon
d
Scen
ario,
2 (two
) disturban
ce
s were
force
d
to the multim
achi
ne
syste
m
, whe
r
e
the me
cha
n
ical to
rqu
e
d
e
viation (
T
1
) was
appli
e
d on M
a
chin
e-2
and
T
2
wa
s ap
plied
on
Machi
n
e
-
3 at
the value of 0.0065 pu a
nd time of
5.0 s. Gra
phi
cal visuali
z
atio
n and num
eri
c
al
values of the
respon
se
s a
r
e discrib
ed in
Figure
s
6, 7 and Tabl
e 2, respe
c
tively.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Dynam
ic Stability Im
provem
ent of Mu
lti
m
achine Power System
s using …
(
A
g
ung
Bu
d
i
Mu
ljo
no
)
1176
Maintena
nce
of the Ma
chine-3 rotor
spe
ed d
e
viation (
3
)
w
a
s
ac
h
i
e
v
ed
w
h
en
the
system
equi
pped by the
proposed PSS.
The
peak ov
ershoot
and
se
ttling time
of this
response
were at th
e
value of
1.
06
10
5
pu
and time
of
5.91
s, re
sp
ectively. The
pea
k ove
r
shoot
responses of conventional PSS
(C
P) and without any
PSS (WP
)
were
obtained at
the
values of
1.48
10
5
a
nd
2.1
3
10
5
pu. And, the settling time re
sp
on
se
s of the
CP
and
WP were
achi
eved at
times of
7.46
and >2
0
s. T
hese respon
ses
are
illu
stra
ting an
d li
stin
g in Fi
gu
re
6(b)
and Tabl
e 2. On the other
hand, sim
u
lat
i
on sh
ows
tha
t
the effect of
the mech
ani
cal torque (
T
2
)
disturban
ce
to the
roto
r
speed
deviatio
n
of M
a
chine
-
2
(
2
) re
sp
ons
e
wa
s v
e
r
y
small.
S
o
,
t
h
is
effect can b
e
negle
c
ted. Th
is re
spo
n
se is sho
w
n in Fig
u
re 6
(
a).
Figure 6. Dynamic st
ability improvem
ent of
rotor speed on the multiple disturbances
Figure 7. Dynamic st
ability improv
em
ent of the rotor angle devia
tion on multiple
disturbances
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 117
0 – 1178
1177
Figure 7
(
a
)
a
nd Ta
ble
2
show the p
e
a
k
ove
r
shoot
and
settling t
i
me of the
ro
tor an
gl
e
deviation M
a
chin
e-2
(
2
) wa
s a
c
hi
eved at the val
ue of
0.341
and
time o
f
5.21 s for t
h
e
proposed PSS, respectively. While, the peak ov
ershoot of the system for conventional PSS
(CP) and wit
hout any PSS (WP) was
at the values
0.342 a
nd
0.67
. And, steady state value
of the
2
was at
0.34
05
. The settling
time for the CP and
WP wa
s at the times of 6.07
and
>20 s.
Finally, imp
r
ovement
re
spon
se
of the
pr
opo
sed
P
SS wa
s
achi
eved at
the
value
of
0.123
a
nd
time of 5.23
s for th
e roto
r an
gle d
e
vi
a
t
ion of Ma
chi
ne-3.
On the
other
han
d, the
pea
k ove
r
sho
o
t of the
CP
and
WP
wa
s
achi
eved
at the valu
es of
0.147
a
nd
0.214
. And,
the
steady
state
value for all
PSS was achieved at
0.1
2
2
. Moreover, the
s
e
ttling time for the
CP
and WP wa
s at
times
of
6.
84
a
nd >2
0 s,
re
spe
c
tive
ly. The
s
e
simul
a
tion results
are ill
ust
r
ated
in
Figure 7(b
)
a
nd Table 2.
Simulation results show
that the proposed PSS is
able to
maintai
n
dynami
c
stability of a
multimachine s
i
gnific
a
ntly in this
res
e
arch
. Where,
the performanc
e
of the propos
e
d PSS is
tested with a
singl
e distu
r
b
ance in First
Scena
rio
an
d
multiple dist
urba
nces in
Secon
d
Scen
ario.
The proposed PSS gives a better perf
o
rmance than the other PSS for a single and mul
t
iple
disturbances.
The
stabilit
y im
provem
ent is achi
eved by r
educi
ng the
peak overshoot and
accele
rating
the settling time of rotor
spe
ed.
Also,
the peak ov
ershoot an
d accele
rating
the
settling
tim
e
of
the angle
deviation are
improve
d
fo
r
respe
c
tive m
a
chi
ne. And,
the pe
rforma
nce
of the proposed PSS are compared to the resp
onses of conventional
PSS and without any P
S
S
t
o
che
ck v
a
lid
it
y
of the
resu
lts.
Table 2. Improvement of a sy
stem
when
the mech
ani
cal torque i
s
force
d
to Machi
ne-3
(
T
2
) at
time 5.0 s
PSS
2
2
Peak overshoot (
P
o)
10
5
(
pu
)
Settling time
(St) (s
)
Po
(
)
St
(s
)
Ss
(
)
WP
5.3
> 20
0.67
> 20
0.3405
CP
0.13
5.94
0.342
6.07
AP
0.10
5.23
0.341
5.21
PSS
3
3
Po
10
5
(
pu
)
St
(s
)
Po
(
)
St
(s
)
Ss
(
)
WP
2.13
> 20
0.214
> 20
0.122
CP
1.48
7.46
0.147
6.84
AP
1.06
5.91
0.123
5.23
7. Conclusio
n
This research is st
ressed on i
m
provement
of dynamic stability a
multimachi
ne system
us
ing ANFIS-based power s
y
s
t
em
s
t
abiliz
er
(proposed PSS). The
proposed PS
S func
tion is
to
provide
ad
ditional stabili
ze
r sign
al
a
s
a torque
da
mpi
ng
com
pone
nt to redu
ce
rotor o
scill
ation.
The rotor oscillation is appeared
when the system is
forced by a
dynamical di
st
ur
bance such as
load
cha
nge
d
or lo
ad flu
c
tu
ation. The A
N
FIS mod
e
l i
s
u
s
ed
in thi
s
re
sea
r
ch be
cause the A
N
FIS
model i
s
m
o
re effective th
an the
Mamd
ani fu
zzy mo
del. The
ANF
I
S-based PS
S is trai
ning
by
the data that obtained by simula
ting conventional PSS. All
the tr
aining processes
are conducted
in off-line
mo
de. Roto
r
spe
ed deviatio
n
and its de
riva
tive are u
s
e
d
as i
nput
s of
the ANFIS-P
S
S
and the
additi
onal
signal st
abilizer of the PSS is ta
ken as an output. Structure of
ANFIS input i
s
built by
Gau
s
sian
mem
b
e
r
ship
fun
c
tion
and it
out
put
is b
u
ilt by Su
geno
fuzzy
orde 1.
Next, the
prop
osed PS
S is applie
d
to a multimachi
ne sy
ste
m
and the
system is forced by a sin
g
le
disturban
ce. The simul
a
tio
n
re
sult
s sho
w
that
re
spo
n
se
s
of the
p
r
opo
se
d PSS
are
bette
r th
an
that responses of the
other PSS.
Where, the peak
over
shoot of the
rotor
speed
deviation of the
proposed PSS is
obtained at the values
of
3.37
10
5
and
1.3
4
10
5
pu for Machin
e-2 a
nd
Machi
n
e
-
3. T
he pe
ak ove
r
sho
o
t of the rotor
an
gle d
e
v
iation is obt
ained
at the values
of
0.
46
and
0.135
for Machine
-
2 and Ma
chi
ne-3, respe
c
tively. The se
ttling time of the rotor
sp
e
ed
deviation i
s
achi
eved
at times of 4.0
1
and
3.98
s,
for
Ma
chine
-
2
and
Ma
ch
ine-3.
And, t
h
e
settling time
of rotor an
gle
deviation i
s
achi
ev
ed at ti
mes
of 4.32
and 4.0
9
s fo
r Ma
chin
e-2
and
Machi
n
e-3, respectively. Furthermore,
the pr
oposed PSS is also able to against mul
t
iple
disturbances.
Wher
e, the proposed PSS gives better resp
onses than that responses of the
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Dynam
ic Stability Im
provem
ent of Mu
lti
m
achine Power System
s using …
(
A
g
ung
Bu
d
i
Mu
ljo
no
)
1178
other PSS
when the
system is forc
ed
by the multiple disturbances
. Some effort
s
shoul
d be done
to improve st
ability of the whol
e multim
achi
ne sy
stem. In the futu
re re
search, the proposed
PSS
should be applied to the other machi
ne i
n
Area II
and III
to test their resp
onses on all system.
Referen
ces
[1]
Z
hang
KF
&
Dai
XZ
. Struct
ural
Ana
l
ysis
of Lar
ge-sca
le
Po
w
e
r S
y
ste
m
s.
Mathe
m
ati
c
a Pro
b
le
m
i
n
Eng.,
Hind
a
w
i
Pub. 201
2.
[2]
Barik S, Mathe
w
AT
.
Desig
n
a
nd C
o
mp
a
r
ison
of Pow
e
r Syst
em Stab
ili
z
e
r
by
Conv
entio
nal
an
d
Robust H
Lo
op Sha
p
p
i
ng T
e
chn
i
qu
e
. Proc
. of ICCPC
T
. 2
014: 12
4-1
29.
[3]
Rou
han, M, Mohamm
adi M, Arefi MM.
Automate
d Mon
i
tori
ng an
d Perfor
ma
nce Assess
me
nt of a Grid
Con
necte
d Sy
nchro
nous
Ge
nerator
Co
nsid
erin
g Pow
e
r S
ystem Sta
b
i
l
i
z
e
r
. Proc. of ICEIA. 201
5: 43
-
48.
[4
]
Ka
ma
la
sa
da
n
S, Sw
ann
GD, Yo
u
s
e
f
ia
n R. A N
o
v
e
l S
ystem-Centric I
n
telli
ge
nt Ad
a
p
tive
Contro
l
Architecture fo
r Po
w
e
r S
y
ste
m
Stab
iliz
er on
Adaptive
Neur
al Net
w
o
r
ks.
IEEE System
J
ournal
. 20
14;
8: 1074-
10
85.
[5]
Phootrak
o
mch
a
i W
,
Jiri
w
i
b
h
a
corn S. Onli
n
e
Critica
l
Cle
a
r
ing T
i
me Estimation Us
in
g an Ad
aptiv
e
Neur
o-F
u
zz
y
I
n
ferenc
e S
y
st
e
m
(ANF
IS).
IJE
PES
. 2015; 73:
170-1
81.
[6]
Mouli
n
LS, da
Silva, APA, El-Sharka
w
i
MA. & Marks II RJ. Support Vect
or Machi
nes fo
r
T
r
ansient
Stabil
i
t
y
An
al
ys
is of Large-sc
al
e Po
w
e
r S
y
ste
m
s.
IEEE
Trans. on Power Syst.
2004; 19 (2)
.
[7]
Ginarsa IM,
Muljo
no AB,
Nrartha IMA,
Cont
ro
lli
ng C
h
aos a
nd V
o
lta
ge C
o
ll
aps
e u
s
ing
La
yer
e
d
Recurre
nt Net
w
o
r
k-bas
ed PI
D-SVC in Po
wer S
y
stems.
Telkom
n
i
ka
. 201
3
;
11(3): 451-4
6
2.
[8]
Ginarsa IM, Soepr
ija
nto A, Purnom
o MH, Cont
rol
lin
g Cha
o
s
and Volta
ge Coll
aps
e usin
g
an
ANF
I
S-
base
d
Comp
os
ite Contro
ller-S
tatic Var Comp
ensator i
n
Po
w
e
r S
y
stems.
IJEPES
. 2013; 4
6
: 79-88.
[9]
Ginarsa IM, So
eprij
anto A, P
u
rnomo MH, S
y
afarud
din, Hi
ya
ma
T
,
Improve
m
ent of T
r
ansient Volta
g
e
Resp
onses
usi
ng an A
dditi
on
al PID-lo
op o
n
an AN
F
I
S-bas
ed Com
posite
Contro
ller-SV
C
(CC-SVC) to
Contro
l Cha
o
s and Vo
ltag
e C
o
lla
pse i
n
Po
w
e
r S
y
stems.
IEEJ T
r
ans. on Pow
e
r and Ener
gy (Sectio
n
B)
. 2011; 13
1(
10): 836-
84
8.
[10]
Ba
w
a
ne N, Ko
thari AG, Koth
ari DP, ANF
I
S base
d
Co
ntrol
and F
a
ult Det
e
ction
of HVD
C
Conv
erter
.
HAIT
Journal o
f
Science an
d
Engi
neer
in
g B
. 2006; 2(5-
6): 673-
689.
[11]
Ginarsa IM, Z
ebu
a O. Stab
ilit
y Improv
em
ent
of S
i
ng
le
Machi
ne
usi
ng ANFIS-PS
S Base
d o
n
F
eedb
ack-li
ne
arizati
on.
Te
l
k
om
ni
ka
. 20
14; 1
2
(2): 315-
32
4.
[12]
Hussei
n
T
&
Shamek
h A.
Adaptiv
e Rul
e
-b
ase F
u
zz
y
Po
w
e
r System Stabil
i
z
e
r for a Multi-
mach
in
e
System
. Proc. of the MED Co
nferenc
e. 201
3
:
1415-1
4
1
9
.
[13]
Kush
w
a
ha
M &
Khar
e
R.
Dyn
a
mic
Sta
b
il
ity Enha
nce
m
ent of
Po
w
e
r Syste
m
Usin
g F
u
zz
y
Lo
gic
Bas
e
d
Power System
Stabili
z
e
r
. Proc
. of Int. Conf. on ICPEC. 201
3
:
213-21
9.
[14]
Shah
B.
Co
mp
arative Study o
f
C
onventi
o
n
a
l
and F
u
zz
y
Bas
ed Pow
e
r System Sta
b
il
i
z
e
r
. Proc. of Int.
Conf. on CSNT
, IEEE. 2013: 547-5
51.
[15]
Padi
ya
r KR.
Po
w
e
r
S
y
stem
D
y
n
a
mic Sta
b
ilit
y
an
d C
o
n
t
rol. Joh
n
Wil
e
y & S
ons
(A
sia) Pte
Ltd
,
Sing
apur
a. 199
4.
[16]
Kund
ur P. Po
w
e
r S
y
stem Sta
b
ilit
y a
nd Co
ntrol. EPRI. McGra
w
-Hi
ll. Ne
w
Y
o
rk. 1994.
[17]
Jang JSR, Su
n CT
&
Mizutani E. Neuro-fu
zz
y
a
nd Soft Comp
uting: A Comp
utation
a
l
Approac
h
t
o
Lear
nin
g
an
d Machi
ne Intel
e
genc
e. Prentic
e-Hal
l
, USA. 1997.
[18]
MAT
L
AB Version 7.9.0.5
29
(2
009
b). T
he Matw
o
r
ks Inc. 200
9.
Evaluation Warning : The document was created with Spire.PDF for Python.