Indonesian J
ournal of Ele
c
trical Engin
eering and
Computer Sci
e
nce
Vol. 2, No. 3,
Jun
e
201
6, pp. 522 ~ 52
9
DOI: 10.115
9
1
/ijeecs.v2.i3.pp52
2-5
2
9
522
Re
cei
v
ed Ma
rch 5, 2
016;
Re
vised
Ma
y 2, 2016; Acce
pted May 1
4
, 2016
Hardwa
re Implementation of Induction Motor using
ANN Controller under Low Spe
e
d Operation
Saran
y
a R
*
, Thangav
e
l
S
Dep
a
rtment of Electrical
and
Electron
ics En
gi
n
eeri
ng, K.S.Ran
gasam
y
C
o
lle
ge of T
e
chnol
og
y,
T
i
ruchengo
de,
T
a
mil Nadu, India
Phon
e No: 82
2
011
01
35
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: saran
y
ar
ama
k
rishn
an9
3@g
m
ail.com
A
b
st
r
a
ct
F
o
r the h
i
g
h
p
e
rformanc
e dr
i
v
es the
artifici
al
n
eur
al
netw
o
rk bas
ed
Ind
u
c
tion
mot
o
r is
prop
osed
.
Durin
g
the
lo
a
d
vari
ation, th
e
perfor
m
a
n
ce
of the In
d
u
ctio
n motor pr
ove
s
to be l
o
w
.
Intellig
ent co
ntro
ller
provi
ded
for c
ontrol
lin
g th
e
spee
d of
in
du
ction
mo
tor e
s
peci
a
lly
w
i
th
hig
h
dy
na
mic
disturb
ances.
An
effective sens
o
r
less strategy
base
d
on
artifi
cial n
eur
al
n
e
tw
ork controller
is deve
l
op
ed t
o
esti
mate roto
r
’
s
positi
on a
nd to
regu
late th
e stator flux u
n
d
e
r low
s
pee
d, he
lps to track the
motor s
p
e
ed
accurate
ly dur
i
n
g
the w
hole
op
er
ating r
egi
on. T
he ov
eral
l co
mbin
ation
of
this
setup is s
i
mul
a
t
ed i
n
the MA
T
L
AB/SIMULINK
platfor
m
. F
i
nal
l
y
an exper
i
m
e
n
tal prototyp
e of the
propos
e
d
drive h
a
s b
een d
e
vel
o
p
e
d
to validate th
e
perfor
m
a
n
ce o
f
Induction M
o
tor and th
e dyna
mic spe
e
d
respo
n
se of
Induction
mo
tor w
i
th propo
sed
control
l
er w
a
s estimated for v
a
rio
u
s spee
d a
nd f
oun
d that the spe
ed ca
n be
contro
lle
d e
ffectively.
Ke
y
w
ords
:
Ind
u
ction
motor, a
r
tificial n
eura
l
n
e
tw
ork, Sensorless Spe
ed C
o
ntrol
Copy
right
©
2016 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
The maj
o
r
Househol
d app
liances in
clu
d
e
cloth
wa
sh
ers, ai
r condi
tioner
s, refri
g
erato
r
s,
vacuum
clea
ners, free
zers, etc. The
s
e
applia
nce
s
h
a
ve [2] traditionally relie
d on cla
s
sic el
e
c
tri
c
motor te
chn
o
l
ogie
s
such a
s
DC m
o
tors,
singl
e
ph
ase
AC ind
u
ctio
n motors. Th
ese
motors a
r
e
typically ope
rated at
con
s
t
ant-spee
d di
rectly fr
om A
C
main
s. Consumers n
o
w d
e
mand
for l
o
wer
energy co
st
s, high
er ef
ficien
cy, better pe
rf
orma
nce, redu
ce
d acou
stic
noise an
d more
conve
n
ient. T
he traditio
nal
techn
o
logie
s
can
not meet
these
dema
nds. Th
e Ind
u
ction m
o
tor
in
these
appli
c
a
t
ions i
s
b
e
co
ming very
co
mmon d
ue to
the features of high
effici
ency, hig
h
flu
x
,
den
sity per u
n
it volume and high po
wer density due
to the absen
ce of field wi
nding in ad
dition
the ab
se
nce
of b
r
u
s
he
s lead
s to
high
relia
bility, low m
a
in
tenan
ce
req
u
irem
ents a
nd
electroma
gne
tic interferen
ce probl
em
s [8].
In this pape
r an effective sen
s
orl
e
ss
strat
egy ba
sed on artifici
al neural net
work is
develop
ed to
estimate
sp
e
ed of i
ndu
ctio
n moto
r a
nd t
o
re
gulate
th
e stato
r
flux u
nder lo
w
spe
ed
and h
e
lp
s to
track the
mot
o
r
spe
ed a
ccurately d
u
rin
g
the whole
o
peratin
g regi
on. Such con
t
ro
l
redu
ce
s the
drive cost,
size an
d maintena
nce
requireme
nts while in
creasi
ng the system
reliability and
robu
stne
ss.
To estimate
the
motor speed, Back
Propa
gation
Network (BP
N
)
algorith
m
is
use
d
in ANN techni
que f
o
r whole
op
eration. T
h
is sen
s
o
r
le
ss
vector
cont
ro
lled
indu
ction mot
o
r d
r
ives
use
d
for hi
gh p
e
rforman
c
e i
n
d
u
strial
appli
c
a
t
ion; it improv
es the
sp
eed
of
the motor un
der lo
w sp
ee
d.
2. Proposed
Metho
d
In these
pro
posed
metho
d
the
thre
e
pha
se i
nput
sup
p
ly is giv
en to
re
ctifie
r
whi
c
h
conve
r
ts AC
to DC an
d is given to the DC lin
k indu
ctor to smooth
en the DC voltage. The
DC
output voltag
e is supplie
d
to the three
p
hase volt
ag
e sou
r
ce
inve
rt
er whi
c
h co
n
v
erts DC
to AC
and th
en
given to th
e in
ductio
n
moto
r. In an
Ar
tificial
Neu
r
al
Network d
e
tects the th
ermal
variation in the stator re
si
stanc
e at different operating
condition. Th
e voltage and
current sig
n
a
l
s
are
determin
ed from
the
output of
the
inverter an
d
is give
n to t
he three p
h
a
s
e to t
w
o
ph
ase
transfo
rmatio
n which
con
s
i
s
t of
Cla
r
ke t
r
an
sform
a
tion
(a
bc to
αβ
)
and Pa
rk T
r
a
n
sformation
(
αβ
to dq).The vol
t
age and current signal
s are given to
ANN controller
whi
c
h cal
c
ul
a
t
e the error a
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 3, Jun
e
2016 : 522
– 529
523
to get the esti
mated rotor position. In trai
ning
data incl
ude input/target data set
s
is used to trai
n
the pro
c
e
s
s.
The blo
c
k dia
g
ram
of the
p
r
opo
se
d sy
stem is
sh
own
in Figu
re 1.
At each
sam
p
le
the neu
ral n
e
twork o
u
tpu
t
is co
mpa
r
e
d
with t
he t
a
rget valu
e
and the
wei
ght co
rrectio
n
is
perfo
rmed
to
minimize the
error bet
wee
n
the
tw
o
val
u
es.
The
out
put related
to
ch
ang
e in
d
u
ty
cycle
whi
c
h i
s
given to
th
e PWM
circui
t. The PWM
gene
rato
r ge
nerate
s
th
e p
l
use
s
whi
c
h
are
given to the
gate d
r
iver.
Whe
n
the firi
ng inp
u
t is
gi
ven to the vo
ltage source
i
n
verter
ba
se
d on
the turni
ng O
N
an
d OF
F i
n
verter
switch, indu
ct
ion motor spe
e
d is
controlled even
un
de
r
l
o
w
spe
ed op
erati
on.
Figure 1. Block di
agram of
the propo
se
d
work
3. Simulation of the Pro
posed Me
th
od
The Simulin
k model of Indu
ction motor d
r
ive us
in
g ANN cont
rolle
r is sho
w
n in Fi
gure 2.
Inductio
n
mo
tor will p
r
o
d
u
c
e diffe
rent t
o
rqu
e
s
wh
en
the roto
r is
runni
ng at di
fferent speed
s
unde
r va
riou
s load
conditio
n
s
and
DC
bu
s voltag
e u
n
d
e
r
ce
rtain tu
rn-on
an
d tu
rn
-off switche
s
.
In
this wo
rk a fe
ed forward n
eural n
e
two
r
k is cr
eate
d
wi
th two neuro
n
s (set spe
e
d
and ch
ang
e
in
spe
ed) in th
e input layer and one ne
uron in the
output layer
with one hid
d
en layer. T
h
e
activation fu
nction
s u
s
ed
for the inp
u
t neuro
n
s
are pu
re lin
ear an
d the
tangent sig
m
oid
activation fun
c
tion is u
s
ed f
o
r output ne
u
r
on.
F
o
r all gi
ven training p
a
tterns the b
a
c
k-p
r
op
agatio
n
training te
ch
nique adj
ust
s
the weight and bia
s
in all con
n
ectin
g
links so th
at the difference
betwe
en the
actual a
nd tar
get output is
minimized.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Hardware Implem
entation of Induction
Motor u
s
ing
ANN Controll
er und
er L
o
w
…
(Sa
r
an
ya R)
524
Figure 2. Simulink
Diag
ram
of Inducti
on
motor d
r
ive using ANN con
t
roller
The te
st sy
stem is sim
u
lat
ed un
de
r vari
ous
low spee
d condition
s.
Figu
re 3
sh
ow th
e
spe
ed re
sp
on
se of the indu
ction moto
r u
s
ing ANN con
t
roller.
Figure 3. Speed wavefo
rm
4
.
Hard
w
a
re
Implementation and disc
ussions
De
sign
of voltage
sou
r
ce in
verter fed i
n
d
u
cti
on
motor
drive u
s
ing A
NN
co
ntrolle
r is al
so
use
d
to elimi
nate the ha
rmonics in the
stator
curren
t and red
u
ce the ripple
co
n
t
ent in gene
rated
electroma
gne
tic torque.
The h
a
rdwa
re re
sult
s fro
m
the p
r
otot
ype such a
s
input a
nd
o
u
tput wavefo
rms are
analysed wit
h
the help
of Digital Storag
e O
sc
ill
oscop
e
. Figu
res
4(a) a
n
d
4(b
)
sh
ow
the
schemati
c
diagra
m
a
nd o
v
erall
Propo
sed Hardwa
re
Setup of
th
e
Indu
ction
motor usi
ng ANN
controlle
r re
spectively
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 3, Jun
e
2016 : 522
– 529
525
(a)
(b)
Figures 4 Experim
ental se
tup, (a) Sche
matic diag
ra
m, (b) overall prop
osed ha
rdwa
re setup
of
the indu
ction
motor
4.1. Voltage Source Inv
e
r
t
er
The main o
b
j
e
ctive of voltage source i
n
verter
s is to
produ
ce a
n
AC output waveform
from a dc p
o
we
r su
pply. This type of wa
veform
s req
u
ired in Adjustabl
e
Speed Drives,
Uninte
rruptibl
e
Powe
r Sup
p
lies, an
d voltage com
pen
sators, whi
c
h
are only a few ap
plicatio
ns.
For si
nu
soid
al ac outp
u
t, the magnitude, fr
equ
en
cy, and pha
se shoul
d
b
e
controllabl
e.
Acco
rdi
ng to
the type of a
c
o
u
tput waveform
,
the
s
e topologi
es ca
n
be co
nsi
d
e
r
ed as
voltag
e
sou
r
ce inve
rt
er, whe
r
e th
e inde
pen
de
ntly controlle
d a
c
outp
ut i
s
a
voltage
waveform.
Ou
tp
u
t
voltage wave
form i
s
sho
w
n in Fi
gure 5.
A 45
W Singl
e
-ph
a
se in
du
ction m
o
tor f
ed by a
volta
g
e
sou
r
ce type inverter
wa
s u
s
ed in the ex
perim
ental
se
tup. The inverter was impl
emented u
s
in
g a
four IGBT po
wer m
odul
e, AT mega microcontrolle
r b
oard of AT3
2
8
microc
ontro
ller pe
rform
s
the
control algo
rit
h
m and ge
ne
rates th
e PWM waveform
s for the IGBT motor d
r
ive inverter
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Hardware Implem
entation of Induction
Motor u
s
ing
ANN Controll
er und
er L
o
w
…
(Sa
r
an
ya R)
526
Figure 5. CRO output
4.2. Pulse Genera
tion
The Pulse wi
dth modulati
on is a mod
u
lated
techni
que that con
f
orm the widt
h of the
pulse, formal
ly the pulse
duratio
n, b
a
se
d on
mo
dulator
sig
n
a
l
informatio
n. Although th
is
modulate
d
techni
que can
be use
d
to enco
de inform
ation for tran
smissio
n
, its main use is to
allow the
co
n
t
rol of the po
wer
su
pplied
to electri
c
al d
e
vice
s, esp
e
c
ially to inerti
al load
su
ch
as
motors, the term duty cy
cl
e describe
s
the propo
rtion
of ON time to the reg
u
la
r interval or
pe
riod
of time.
Figure 6. Switchin
g plu
s
e
s
Wavefo
rm
The pul
se
g
eneration of
the voltage
sou
r
ce can
be obtai
ne
d by usi
ng
ATmega
microcontroll
er
T
he Figu
re
6 sho
w
s
the
input sw
it
chin
g plu
s
e
s
. Ba
sed o
n
the
sp
eed va
riation
of
the indu
ction
motor cl
osed
loop c
ontrol t
he pul
se can
be varied.
4.2. Driv
er Circuit
The indu
ctio
n motor is p
o
we
red
with the
help of d
r
iver ci
rcuit. The output o
f
voltage
sou
r
ce inve
rter i
s
conve
r
ted into
DC
whi
c
h i
s
furt
her
given a
s
input to the
driver
ci
rcuit. The
output of the
driver
circuit f
o
r
two
differe
nt operating
spe
ed of t
he
motor i
s
in trape
zoid
al sh
ape
whi
c
h is
sho
w
n in Figu
re
7(a
)
and 7
(
b
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 3, Jun
e
2016 : 522
– 529
527
(a)
(b)
Figures 7. Th
e output wav
e
form for o
p
e
r
ating spee
d of the motor at frequen
cy
(a) 7
57.6 H
z
, (b) 1
63.9 H
z
In the above
waveform, the frequ
en
cy d
enoted i
s
for
one full cy
cle.
The ind
u
ctio
n motor
is o
perated i
n
varia
b
le
op
erating
spee
d conditio
n
.
So the frequ
ency
of the
motor i
s
va
ri
ed for
different ope
rating sp
eed o
f
the motor.
4.3
.
DC Link
Voltage Wav
e
form
DC lin
k exits between
a
rectifier an
d i
n
ve
rter.
On
one
end, th
e
utility con
n
e
c
tion i
s
rectified
into
a high
voltag
e DC. On
the
other end,
t
hat DC i
s
swi
t
ched
to ge
n
e
rate
a ne
w
AC
power waveform. It’s a link be
cau
s
e it
conne
cts the i
nput and o
u
tp
ut stage
s.
The te
rm
“DC link” i
s
also
use
d
to
de
scribe the
de
co
u
p
ling
ca
pa
cito
r in
the
DC li
nk. T
he
swit
chin
g net
work
on the
o
u
tput sid
e
ge
nerate
s
ve
ry large
tran
sie
n
ts at the
switching freque
n
cy.
The
DC li
nk
cap
a
cito
r h
e
l
p
s to
ke
ep th
ese t
r
an
sie
n
ts fro
m
radiati
n
g ba
ck to th
e input. Th
e
DC
link voltage
waveform from
the SMPS is
sho
w
n in the
Figure 8.
Figure 8. Output of DC Lin
k
Voltage
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Hardware Implem
entation of Induction
Mo
tor u
s
ing
ANN Controll
er und
er L
o
w
…
(Sa
r
an
ya R)
528
Figure 9. View of Experim
ental Setup
PWM wavefo
rm from mi
croco
n
trolle
r is
then
amplifie
d with op
en
collecto
r
opto
cou
p
ler
and fed to th
e DC chop
pe
r thro
ugh
an i
s
olato
r
an
d d
r
iver
chip. T
h
e DC chop
pe
r outp
ut is giv
en
to the armatu
re of the
ind
u
ction
motor.
The vo
ltag
e
tran
sform
e
r
con
n
e
c
ted to
the moto
r shaft
gives a
DC
voltage prop
ortional to th
e sp
eed, an
d
this DC voltage i
s
fed to
the ADC
of the
microcontroll
er. The Experimental se
tu
p
used i
s
given
in Figure 9.
5. Sy
stem under Various
Opera
t
ing Speed Conditi
on
The te
st sy
stem is sim
u
la
ted un
der va
ri
ou
s o
perating
conditio
n, while
the In
ductio
n
motor is working unde
r co
n
s
tant load co
ndition, the
current dra
w
n
from the input
supply is 0.8
A.
The expe
rime
ntal respon
se
s of an Induct
i
on motor
for
a step ch
ang
e in refere
nce spe
ed from
to
rated speed with ANN
c
o
ntroller .
10(a
)
10(b
)
Figure 10(a)
and 10
(b
). Co
ntrolling the S
peed u
nde
r 1
1
79 rp
m and
826 rp
m re
sp
ectively
Figures
10
(a) and
10(b)
show th
e
cont
rolling
the
sp
eed of th
e in
ductio
n
moto
r u
s
ing
ANN controll
er
unde
r
117
9rpm
an
d 8
2
6
rpm
respect
i
vely, While t
he Ind
u
ctio
n
motor i
s
wo
rking
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 3, Jun
e
2016 : 522
– 529
529
unde
r loadi
ng
(LOW) load
con
d
ition, the current dr
awn from the input supply is
0.849A. At th
at
time the spee
d maintai
ned
as 8
2
4
r
pm
(A
ctual Sp
eed
)
for the
826
rp
m (Set Spe
e
d
) a
s
sho
w
n
i
n
the Figu
re 1
0
(
b). T
he d
e
si
gned
co
ntroll
ers were
i
m
pl
emented
by u
s
ing
an Atmel
microcontroller.
The
rating
s
of indu
ction
motor
used
are
220/2
3
0
V
, 45
watts,
and
24
00
rpm. T
he A
N
N
controlle
r’s p
e
rform
a
n
c
e
wa
s teste
d
with the
ind
u
c
tion moto
r. The result are com
pared
and
tabulated a
s
sho
w
n in the
Table 1.
Table 1. Anal
ysis of Simul
a
tion Re
sult
Char
ecter
i
stics
Simulation Resul
t
Har
d
w
a
r
e
Result
Speed
Phase Voltage
Linearl
y
varied a
nd settled do
w
n
slow
l
y
to th
e refe
rence speed
(870rpm
) (0
.065
ms)
The Phase Volta
ge Waveform is
exactl
y
in tra
pezoidal shape
The speed of t
h
e
motor is kept
varies under lo
w
load
condition(861rp
m
) b
y
controlling DC link voltage w
i
th
the help of ANN
Control
l
e
r(BP
N
) ,
mi
ni
mum
oscillation occurs
The Phase Volta
ge Waveform
is in trapezoidal
shape
w
i
th
minimum distortions
6. Conclusio
n
In this p
ape
r, the propo
se
d metho
d
Vo
ltage Sou
r
ce
Inverter fe
d
IM drive u
s
in
g ANN
controlle
r u
n
der lo
w
spe
e
d
op
eration
condition
is ex
perim
entally i
m
pleme
n
ted
usin
g AT
me
ga
microcontroll
er bo
ard of A
T
Mega 32
8
for a 45
W si
ngle ph
ase Indu
ction mot
o
r. The p
r
op
ose
d
scheme
has sho
w
n satisfacto
ry perf
o
rma
n
ce
for wide rang
e of spee
d
variation. The
comp
arative analysi
s
of the simulate
d a
nd exper
i
m
e
n
tal re
sults
shows t
he effe
ctivene
ss of t
h
e
prop
osed d
r
i
v
e system. T
he propo
se
d
scheme i
s
re
comm
end
ed
solutio
n
for t
he ap
plicable
of
low power Induc
tion motor. In the future work
,
re
ne
wabl
e en
ergy
like Sol
a
r, F
uel cell
can
be
use
d
a
s
the rene
wable
en
ergy sou
r
ce for this
sy
ste
m
whi
c
h i
s
u
s
eful to redu
ce the d
e
ma
nd of
electri
c
ity. It a
l
so re
du
ce
s the pollution
a
nd gre
enh
ou
se effect.
Referen
ces
[1]
KS Naren
d
ra
and K Parth
a
s
a
rath
y. “Identifi
c
ati
on a
nd C
o
ntrol of D
y
na
mic S
y
stems
usin
g Neur
al
Net
w
orks”.
IEEE Trans. Neural Networks
. 1990; 1: 4-27.
[2]
Z
u
rada JM. “
Introducti
on to Artificial N
eur
al S
ystems
”. Mum
bai: Jaic
o Pub
l
i
s
hin
g
Hous
e. 1
992.
[3]
Zak
y
MS. “Stabilit
y Ana
l
ysis
of Speed a
nd
Stator
Resista
n
ce Estimators
for Sensorles
s
Induction
Motor Drives”.
IEEE Trans. In
d. Electron
. 20
12; 59: 85
8–8
7
0
.
[4]
Gadou
e SM, G
i
ao
uris
D a
n
d
F
i
nch JW
. “MR
AS Sen
o
rl
ess
Vector C
ontrol
of an
Ind
u
ction
Motor
usin
g
Ne
w
Sl
idi
ng-M
ode
and F
u
zz
y-L
o
g
i
c Ada
p
ta
tion Mech
an
is
ms”.
IEEE Tra
n
s. on Energy
Convers
i
on
.
201
0; 25(2): 39
4-40
2.
[5]
Ibrahim Moh
d
Alsof
y
an
i, Nik Rumzi Nik Idris
.
“S
imple F
l
u
x
Reg
u
lati
on for Improvin
g State Estimati
o
n
at Ver
y
L
o
w
an
d Z
e
ro Spee
d
of a Spee
d Se
nsor
l
e
ss Direct
T
o
rque Co
ntro
l of Inductio
n
Motor”.
IEEE
T
r
ansactio
n
s o
n
Pow
e
r Electronics
. 20
15; 18
(2).
[6]
F Balaabjerg, M Lise
rre
an
d
K Ma. “Po
w
er
Electron
ics Co
nverters for W
i
nd T
u
rbine S
ystem”.
IEEE
T
r
ans. Ind. Appl
. 201
2; 48(2):
708-7
19.
[7]
Bazzi AM, Bu
yukde
g
irme
nci
VT
, Krein PT
. “S
y
s
tem-L
eve
l
Po
w
e
r
Loss
S
ensitivit
y t
o
Va
rious
Co
ntro
l
Varia
b
les
in V
e
ctor-Co
n
troll
e
d Inducti
on M
o
tor Driv
es”.
IEEE Trans. Ind. Appl
.
201
3;
49(3): 1
3
6
7
-
137
3.
[8]
Khob
ara
gad
e
T
,
Barve A. Enha
ncem
ent
of po
w
e
r s
y
stem stabil
i
t
y
u
s
ing fuzz
y l
o
g
i
c control
l
er.
Internatio
na
l Journ
a
l of Pow
e
r El
ectronics a
nd Driv
e System (IJPEDS)
. 2012: 2 (4): 38
9
-
401.
[9]
Buja
GS a
n
d
T
odesco F
.
“Neura
l
net
w
o
rk impl
eme
n
ta
tion
of a
fuz
z
y
lo
gic c
ontr
o
ller”.
IEEE
T
r
ansactio
n
s o
n
Industria
l Ele
c
tronics
. 199
4; 41(6):
66
3–
665
.
Evaluation Warning : The document was created with Spire.PDF for Python.