Documente Academic
Documente Profesional
Documente Cultură
DOI 10.17148/IJIREEICE.2015.3105
23
II. MODELLING
A. Induction Motor
The three-phase squirrel cage induction motor
mathematical equations in synchronous rotating reference
frames are as follows [1][3]:
Ve R ie p e e
ds
s ds
ds
e qs
(1)
Ve R ie p e e
qs
s qs
qs
e ds
(2)
0 R ie p e ( ) e
r dr
dr
e r qr
(3)
0 R ie p e (w w ) e
r qr
qr
e
r dr
(4)
Where,
e L ie L ie
ds
s ds
m dr
e L ie L ie
qs
s qs
m qr
e L ie L ie
dr
r dr
m ds
e L ie L ie
qr
r qr
m qs
(5)
(6)
(7)
(8)
S i1
Si 2
Si 3
Si 4
ON
ON
OFF
OFF
Vdc / 2
OFF
ON
ON
OFF
OFF
OFF
ON
ON
Vdc / 2
3P
T
L (i e i e i e i e )
e 2 2 m qs dr ds qr
d
w r
r
dt
dw
r B w T
T j
e
m dt
m r
l
Copyright to IJIREEICE
(9)
(10)
(11)
DOI 10.17148/IJIREEICE.2015.3105
24
Hysteresis
Band
ia*+h
ia*+
ia*
ia*-
ia*- Vdc/2
Dead Zone
Vdc/2
-Vdc/2
-Vdc/2
is
Vs
Vqse
s
qs
Vdc
Vdss
sl
s
qs
Vdse
iqse
dre r
e
ds
(q-d, e)
Synchronous
reference frame
(q-d, r)
switching. A dead zone is necessary in the hysteresis
idss
Rotor reference
band h, to avoid switching towards two-level scheme,
frame
because of finite sampling rate of error. Without the dead
(q-d, s)
Stator Reference
zone, when the error becomes zero and is not detected, the
frame
opposite polarity of forcing function follows, resulting in a
two-level scheme. However, the introduction of dead zone
Figure 4: Phasor diagram of Indirect Vector Control principle
increases the tracking error and has to be chosen to a
For decoupling control 0 or p 0 and
minimum value, depending on the best sampling speed
qr
qr
that can be achieved [15].
.Substituting the above condition in equations
dr
If U represents the input state to be applied, e represents r
error (ia* ia ) and ce represents the change in error then (3), (4), (7) and (8).
from Figure 2 and Figure 3, the switching logic is
R L ie
r m qs
governed by equation (12)
(14)
sl
L
If e 0 then
r r
U 1 for e h
3 P Lm
(15)
T
ie
U 0 for e
e 2 2 L r qs
r
U 0 for e h and ce 0
2 2 Lr Te
U 1 for e h and ce 0
(16)
ie
qs
3PL
Else if e 0 then
m r
U -1 for e -h
L
1
(17)
ie
[ r p ]
U 0 for e -
r
ds
U 0 for - e - h and ce 0
L
m
R
r
m r
inverter can be obtained.
*
1 * Lr
ie
[
p * ]
ds L
r R
r
m
r
(20)
DOI 10.17148/IJIREEICE.2015.3105
25
*
*
e
r
i
ds L
m
Bv K i i
(23)
X AX
E. Speed Estimator
s
s s
The proposed scheme includes a Luenberger Observer for
i s
ds
the speed estimation from stator voltage and current. The
Where,
i CX
Luenberger Observer belongs to the group of closed loop
s
s
iqs
observers. It is a deterministic type of observer because it
is based on a deterministic model of the system [17].This
0 0
observer can reconstruct the state of a system observable C I 0 and 0
0 0
from the measurement of inputs and outputs. It is used
when all or part of the state vector cannot be measured. It
allows the estimation of unknown parameters or variables The symbol ^ represents the estimated values and K is the
observer gain matrix which is to be decided so that the
of a system.
The state equations of an Induction motor in stationary above equation can be stable.
V
V
reference frame are given by.
Transformation
s
ds
abc
(a,b,c) (d,q,s)
X AX Bv
s
(22)
V qss
idss , iqss
i ds , i qs
Where,
i s
i
ds
s
X , i ,
s
s
r
iqs
A
11
A
A
21
s
dr
r
,
s
qr
Particle Swarm
Optimised
Fuzzy Logic
Controller
A
12
A
22
Transformation
(a,b,c) (d,q,s)
iabc
dr , qr
s
s
eids qr eiqs dr
eids , eiqs
G a in
K
R
(1 )
A s
I I ,
11
L
T
s
r
L
m 1 I J 1 I J ,
A
12 L L T
r
r
T
s r r
L
A m
21 T
r
B
B 1 ,
0
I,
A I J ,
22
r
T
r
v s
ds
v
,
s
s
vqs
L2
1- m ,
L L
s r
s 1 ,
L
T
r
s
e A KC e AX
Where,
1
B
I
1 L
s
0 -1
J=
,
1 0
L
T r,
r R
r
e X - X ,
(24)
0 - J
r
A A - A
,
0 r J
r
r
r
Now lets define the Lyapunovs function candidate:
V eT e
Here,
r r
(25)
r is
L
m
L L
s r
Thus if the speed signal r is known in the matrix A then Lyapunov function V is negative definite. The time
derivative of V becomes
the fluxes and currents can be solved from the state
equations. However, r is not correct, there will be a
deviation between the estimated states and actual states.
So the estimated currents are compared with the actual
machine terminal currents and the errors are injected
Copyright to IJIREEICE
T 2 r r
V eT A KC T A KC e AT X T e AXe
(26)
DOI 10.17148/IJIREEICE.2015.3105
26
e s e s
ids qr iqs dr
L
2
m
r
2
r
r
L
L
r
S
P I K
1
o
P I A KC
o
L
m I K
2
T
0
1
P
I J
o T
r
r
I J
r
T
r
(33)
Where, e i s i s and e i s is
ids ds ds
iqs qs qs
a - c K K b 0
(35)
1
2
1 k a b 1 k 1 I J (36)
s
s
K
e e
1
p
p
r
T
r
ids qr iqs dr
r
1 k 2
1 k
p
p
K
a c
a b
2
Or
(28)
s e s dt
r
ids qr iqs dr
1 k
2
r
are selected as:
(29) k k 1
P k P
o
p m
2 p r
Where, k p 1 it can be achieved by defining the observer
matrix K in a special form
k I k J K
1
2 1
(30)
K
k3 I k4 J K 2
To determine the Eigen values of the matrix A,
1
P I
I J
r
T
m
r
P IA
0 (31)
m
L
1
m I P
I J
m T
r
T
r
r
To simplify the above equation, substitute
1
L
a I, b
I J and c= m I
r
T
T
r
1 k 2
L
p
m
k
3
1 k p
(38)
1 k
p
k
4
r
Optimised
Fuzzy Controller
780V
Estimation
of
Particle Swarm
T e*
*
sl
+
+
i de s*
iqse*
Transformation
(q, d , e) (a, b, c)
Controller
Level
Inverter Vabc
iabc
iabc
Luenberger Observer
For Speed Estimation
Vabc
Copyright to IJIREEICE
DOI 10.17148/IJIREEICE.2015.3105
27
e
Normalizer
Fuzzifier
ce
du
Inference
Mechanism
De-Fuzzifier
NS
a2
a1
Knowledge
Based rule
NB
New Fuzzy
Parameters
By PSO
B. Fuzzifier / De-fuzzifier
The fuzzifier processes the crisp input values (e, ce) and
convert them into fuzzy values. Also the fuzzy values
obtained in fuzzy inference mechanism are converted to
crisp output du value by a de-fuzzifier. Here a triangular
fuzzy membership function is defined for each input and
output values by seven clusters. For seven clusters in the
membership functions, seven linguistic variables are
defined as: Negative Big (NB), Negative Medium (NM),
Negative Small (NS), Zero (Z), Positive Small (PS),
Positive Medium (PM), Positive Big (PB). Figure 8 shows
the membership functions used to fuzzify two input values
(e, ce) and de-fuzzify output du of the realized fuzzy
controller. The peak or bottom points of the membership
functions to be tuned are a1 and a2 for error e, b1 and b2 for
change in error ce and c1 and c2 for output du. Therefore
the design of fuzzy controller require the optimization of
nine parameters (n1,n2,n3,a1,a2,b1,b2,c1,c2).
The centre of gravity or centroid method is used for defuzzification. As a result the control increment is obtained
by the equation [20]
PB
PS
PM
NS
b2
b1
NB
NM
NS
c2
c1
(ce)
a1
a2
PS
PM
PB
(du)
b1
b2
PS
PM
c1
c2
PB
(e)
Z
m
d A
i
i
du i 1
m
A
i
i 1
th
NM
A. Normalizer / De-normalizer
In closed loop control system the use of error (e) and the
change in error (ce) as controller input is a universal
approach. Therefore the fuzzy logic controller has two
inputs, error and change in error (e, ce) and one output du
as shown in Figure 7.Two normalization parameters (n1,
n2) for inputs (e, ce) and one de-normalization parameter
(n3) for output du are defined. In normalization process the
input variables are scaled in the range of (-1, +1) and in
de-normalization process the output values of fuzzy
controller are converted to a value depending on the
terminal control element. The determination of
normalization and de-normalization parameters of fuzzy
controller is important for system stability.
Copyright to IJIREEICE
NM
De-Normalizer
Here
NB
i fuzzy set.
DOI 10.17148/IJIREEICE.2015.3105
49
(du)
(41)
28
i
i
1 1
i
i
+C r ( n) G
x ( n)
2 2
best
i
(43)
(44)
(45)
(42)
min n
( n)
max
max
n
max
0 (C C ) 4
1
2
(C C )
1
2 1 ( n) 1
2
ce (t ) e (t ) e (t 1)
c
c
c
(48)
(49)
(51)
t
F dt
e
r
r
0
(52)
DOI 10.17148/IJIREEICE.2015.3105
29
Start
Inputs: Max. no. of iterations, Population size, no.
of variables, Cognitive acceleration, social
acceleration, Max. inertia weight, Min. inertia
weight, Upper boundary and Lower boundary of
each variable, Constraints and Termination criteria.
No
nc1
ne1
0.204230794614153
nc2
nc3
6.366601440561e04
0.538296104367619
3.474559679239366
ne2
ne3
0:372717279683677
0:958405412315019
ac1
0.035691605541264
ae1
0:479261688178227
ac2
bc1
0.700814506194795
0.053227790847909
ae2
be1
0:904836175738316
0:142646332443648
bc2
cc1
cc2
0.793644125939212
0.050531304972714
0.695057495825054
be2
ce1
ce2
0:768274946372518
0:460490418982960
0:533696655366175
Yes
Is
Termination criteria reached?
Optimal
Values
Stop
No
N=N+1
No
Satisfy Constraints ?
x 10
Mean Score
Best Score
1.74
1.72
1.7
1.68
1.66
1.64
0
Parameter
Symbol
Value
Rated Power
Prated
50 Hp
480 Volt
Rated Frequency
50 Hz
Pair of Poles
Stator Resistance
Rs
0.087
Rotor Resistance
Rr
0.22
Stator Inductance
Ls
0.8 mH
Rotor Inductance
Lr
0.8 mH
Mutual Inductance
Lm
34.7 mH
Moment of Inertia
1.662 kg.m2
x 10
30
40
Mean Score
Best Score
10
0
0
10
15
20
25
Generation
30
35
40
45
50
DOM
PM
PB
1 NB
-0.8
-0.6
-0.4
-0.2
NM
0
Error (e)
0.2
0.4
0.6
NS Z PS
0.8
PM
PB
0.8
0.5
0
-1
DOM
NS Z PS
NM
1 NB
0.5
0
-1
20
Generation
Score
DOM
Rated Voltage
10
15
Copyright to IJIREEICE
1.76
Score
-0.8
1 NB
-0.6
-0.4
-0.2
0
0.2
Change in Error (ce)
NM
NS Z
0.4
0.6
PS
PM
PB
0.5
0
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
Output (du)
0.4
0.6
0.8
DOI 10.17148/IJIREEICE.2015.3105
30
DOM
0
-1
NB
1
NS
PS
PM
PB
120.3
-0.8
-0.6
-0.4
-0.2
NM
NS
0
0.2
Error (e)
Z
PS
0.4
0.6
0.8
PM
1
PB
-0.8
-0.6
-0.4
NM NS
NB
1
-0.2
0
0.2
Change in Error (ce)
Z
0.4
-0.2
0.4
0.6
0.8
PS PM
1
PB
0.5
0
-1
-0.8
-0.6
-0.4
0
0.2
Output (du)
0.6
0.8
0
2.05
2.1
2.15
2.2
2.25
Time(sec)
2.3
2.25
Time(sec)
2.3
2.35
2.4
2.45
2.5
Torque Developed(Nm)
0
2.05
2.1
2.15
2.2
2.35
2.4
2.45
2.5
40
0
-40
2
2.05
2.1
2.15
2.2
2.25
Time(sec)
2.3
2.35
2.4
2.45
1.5
2.5
Time(sec)
3.5
4.5
Torque Developed
Load Torque
200
100
0
0
0.5
1.5
2.5
Time(sec)
3.5
4.5
100
50
0
0
1000
-1000
2
0.5
300
500
-500
2
120
Speed(Rad./Sec)
Line Voltage(V)
Pole Voltage(V)
Stator Currents (A))
Motor Speed
Estimated Speed
Reference Speed
120.2
119.8
0
0.5
0
-1
DOM
NM
Speed(Rad./Sec)
NB
1
0.5
Torque Developed(Nm)
DOM
INTERNATIONAL JOURNAL OF INNOVATIVE RESEARCH IN ELECTRICAL, ELECTRONICS, INSTRUMENTATION AND CONTROL ENGINEERING
Vol. 3, Issue 1, January 2015
0.5
1.5
Motor Speed
Estimated Speed
Reference Speed
2.5
3
3.5
Time(sec)
4.5
Torque Developed
Load Torque
200
0
-200
0
0.5
2.5
1.5
2.5
Time(sec)
3.5
4.5
Figure 14 shows the pole voltage, line voltage and the line
currents (stator line currents) of the three-level inverter
under steady state condition. The line currents are
sinusoidal with almost negligible ripple. Figure 15 shows
the speed tracking performance of the motor following a
trapezoidal speed reference. The speed tracking
experiment is on no load condition. The motor speed
almost tracks the reference speed in both the direction.
60
Motor Speed
Estimated Speed
Reference Speed
Speed(Rad./Sec)
40
REFERENCES
20
-20
-60
0
3
Time(sec)
Speed(Rad./Sec)
0.5
1.5
Time(sec)
Motor Speed
Estimated Speed
Reference Speed
2.5
3
0.5
1.5
Time(sec)
2.5
119.9
119.8
0
400
200
0
0
-40
Torque Developed(Nm)
VI. CONCLUSIONS
The two optimal fuzzy logic controllers have been
designed off-line using techniques of Particle Swarm
Optimization for sensor less vector control of multilevel
inverter fed Induction motor. It achieves good estimation
of motor speed, pursuit of reference speed, starting
without overshoot and rapid rejection of disturbances with
a low drop-out speed. From simulation results, it testifies
that this method is not only robust, but also can improve
dynamic performance of the system.
DOI 10.17148/IJIREEICE.2015.3105
31
Copyright to IJIREEICE
DOI 10.17148/IJIREEICE.2015.3105
32