Faculty of Electrical Engineering

Universiti Teknologi Malaysia

VOL. 11, NO. 1, 2009, 38-43
ELEKTRIKA
http://fke.utm.my/elektrika

38
A Simple Fuzzy Modeling of Permanent Magnet
Synchronous Generator
N. Selvaganesan
1*
and R. Saraswathy Ramya
2

1
Faculty in Avionics, Indian Institute of Space science and Technology(IIST), Department of Space, Govt. of India,
Thiruvananthapuram 695022, India.
2
Application Development Group, Collabera Solutions Pvt. Ltd, Bangalore-560095, India.

*
Corresponding author: n_selvag@rediffmail.com, Tel: +91 471 2563484, Fax: +91 471 2564806
Abstract: This paper presents the design methodology of fuzzy based modeling of permanent magnet synchronous generator
system. An analytical model of the system is presented in the nonlinear state space form based on fundamental physical laws.
The fuzzy based modeling scheme for Permanent Magnet Synchronous Generator is developed using the general Takagi-
Sugeno fuzzy model. The feasibility of the proposed scheme for Permanent Magnet Synchronous Generator is demonstrated
using MATLAB for load disturbances and the obtained responses are compared with mathematical model responses.
Keywords: T-S Fuzzy model, permanent magnet synchronous motor, rotor reference frame, fuzzy system.
1. INTRODUCTION
Permanent Magnet Synchronous Generators (PMSGs) are
receiving significant attention from industries for the last
two decades. They have numerous advantages over the
machines which are conventionally used. Current
research in the design of the PMSG indicates that it has
high torque to current ratio, large power to weight ratio,
high efficiency, high power factor and robustness [1].
Currently, there is much interest in using brushless
electronically commutated servo machines in high
performance electromechanical systems and the
application of neodymium-iron-boron (Nd
2
Fe
14
B) and
samarium cobalt (Sm
1
Co
5
and Sm
2
Co
17
) rare-earth
magnets results in high torque and power density,
efficiency and controllability, versatility and flexibility,
simplicity and ruggedness, reliability and cost, weight-to-
torque and weight-to-power ratios, better starting
capabilities [2]-[5].
Building models of reality is a central topic in many
disciplines of engineering and science. Models can be
used for simulations, analysis of the system's behavior
and for a better understanding of the underlying physical
mechanisms in the system. In control engineering, a
model of the plant can be used to design a feedback
controller or to predict the future plant behavior in order
to calculate optimal control actions. The mathematical
modeling of PMSG is discussed by researchers in various
literatures [1]-[6].
Recent advances in the theory of fuzzy modeling and a
number of successful real-world applications show
that fuzzy models can be efficiently applied to complex
nonlinear systems intractable with standard linear
methods. The idea of fuzzy modeling was first proposed
by Zadeh [7] and has subsequently been pursued by many
others for more than two decades. The fuzzy modeling for
complex processes is regarded as one of the key problems
in fuzzy systems research [8] [9].
In the field of fuzzy modeling, the Takagi - Sugeno (T-
S) fuzzy model [10]-[15] has been used to approximate
accurately the dynamics of complex systems. Besides the
capability of modeling nonlinear systems, there are other
properties that make fuzzy models interesting not only
theoretically but also for the industrial practice. Few
researches [16]-[17] have attempted in making a fuzzy
model for permanent magnet synchronous motor drive.
But the fuzzy modeling of PMSG is largely unexplored.
This paper focuses the attention on T-S fuzzy modeling
of PMSG drive. The proposed fuzzy model for PMSG is
simulated with load disturbances and the responses are
compared with the mathematical model responses.
The rest of the paper is described in four sections.
Section 2 explains the mathematical model of PMSG
drive. Section 3 deals with proposed T-S fuzzy model for
PMSG drive. Section 4 describes the comparison between
the mathematical model and the fuzzy model under load
disturbances. Finally the conclusion is narrated in section
5.

2. THE PMSG SYSTEM MODEL

The dynamic characteristics of a PMSG can be modeled
based on the d-q axis and the differential equations that
describe the circuitry and torsional-mechanical dynamics
[1]-[5]. Consider the generation system when the
permanent magnet DC motor is used as prime mover
whose differential equations are as follows.

a a a
a r a
a a a
di r k 1
= i - + u
dt L L L
(1)

r a rm m m
a qs rm
k d 3P B
= i - i -
dt J 4J J
(2)
rm
rm
d
=
dt
(3)
N. SELVAGANESAN, R. SARASWATHY RAMYA / ELEKTRIKA, 11(1), 2009, 38-43
39
where, r
a
, L
a
, i
a
, k
a
are armature resistance, armature
inductance, armature current , armature constant
respectively.
r
,
rm
and
rm
are angular velocity,
mechanical angular velocity and mechanical angular
displacement respectively. P is number of poles.
m
is
flux linkages. B
m
and J are moment of inertia and viscous
friction co-efficient respectively.
r
qs
i
is the q-axis stator
current component. The transient dynamics of generator
must be integrated with prime mover equations. In the
rotor reference frame, mathematical model of three phase
PMSG [1] with resistive load can be expressed as follows

r
qs r r s L m
qs rm ds rm
ls ls
m m
di
r +R P P
=- i + - i
3 3
dt 2 2
L + L L + L
2 2
(4)
r
r r ds s L
ds qs rm
ls
m
di r +R P
=- i + i
3
dt 2
L + L
2
(5)

Where,
r
ds
i
is d-axis stator current component R
L
and r
s
are
load resistance and stator resistance respectively. L
ls
and
m L are leakage and magnetizing inductance
respectively. The q-axis voltage and d-axis voltage are
given [1] as follows.

r r
qs s ss qs r ss ds r m
d
u r L i L i
dt

= + +


(6)
r r
ds s ss ds r ss qs
d
u r L i L i
dt

= + +


(7)

Where, u
ds
, u
qs
are q- axis stator voltage component, d-
axis stator voltage component respectively.
ls
m
3
L + L
2
is
L
ss
which is the self inductance of stator winding. By
means of Parks transformation the generated three
current and voltages are obtained.

3. T-S FUZZY MODEL
Fuzzy modeling is derived based on T-S fuzzy model. It
is derived by utilizing the concept of Parallel Distributed
Compensation. The design procedure is conceptually
simple and natural. It begins with representing a non-
linear plant based on T-S fuzzy model. The fuzzy model
proposed by Takagi and Sugeno is described by fuzzy
IF-THEN rules, which represent local linear input-output
relations of a nonlinear system. The main feature of a T-S
fuzzy model is to express the local dynamics of each
fuzzy implication (rule) by a linear system model. The
overall fuzzy model of the system is achieved by fuzzy
blending of the linear system models. Many nonlinear
dynamic systems can be represented by T-S fuzzy model
and are proved as universals approximators [13]-[15].
Consider a general system as follows:
sx(t)=f(x(t))
y(t)=h(x(t)) (8)
Where, x R
n
is the state vector; sx(t) denotes x (t)
.

and x(t+1) in continuous-time and discrete-time systems
respectively; y R
m
is the systems output; f() and h()
are nonlinear functions with appropriate dimensions. A
fuzzy representation of (8) is composed of the following
rules:
IF z
1
(t) is F
1i
and .. and z
g
(t) is F
gi
THEN
sx(t) = A
i
x(t)+b
i
y(t)= C
i
x(t), i = 1, 2, ., r (9)

Where, z
1
(t) ~ z
g
(t) are the premise variables which
consists of states of the system; F
ji
(j = 1, 2, , g) are
fuzzy sets, r is the number of fuzzy rules, A
i
and C
i
the
are system and output matrices with appropriate
dimensions and b
i
R
n
denotes the constant bias term,
which is generated by the exact fuzzy modeling
procedure.
Using the singleton fuzzifier, product fuzzy inference
and weighted average defuzzifier [14], the final output of
the fuzzy systems is inferred as follows:
sx(t)=
r
i i i
i = 1
(z(t))(A x(t)+b )

y(t)=
r
i i
i = 1
(z(t))(C x(t))

(10)

Where z(t) = [z
1
(t) z
2
(t) . z
g
(t)]
T
, and
i
i r
i
i = 1
(z(t))
(z(t))=
(z(t))

with
g
i ji j
j = 1
(z(t))= F (z (t))

note that
r
i
i = 1
(z(t)) = 1

for all t, where

i
(z(t)) 0
for i = 1, 2, , r.

Considering equations (4) and (5), the fuzzy
membership functions are specified appropriately in
premise parts, associated entries of matrices A
i
, C
i
and b
i

in the consequence parts are represented by a TS fuzzy
model. From the investigation of a large class of
continuous-time and discrete-time systems, it is found
that nonlinear terms have a common variable or depend
only on one variable. If we take this as the premise
variable of fuzzy rules, a simple fuzzy dynamic model
can be obtained.
From the equations (4) and (5) X
1
and X
2
are the state
variables of quadrature-axis current and direct-axis
current

respectively and the speed is chosen as the
premise variable. The S
1
and S
2
denotes the maximum and
the minimum values of speed [0 - 444] rad/sec. The
discretised equations of (4) and (5) are written as follows

X
1
(k +1) = t j-

s
+R
L
L
ss
X
1
(k) -
Po
rm
2
X
2
(k) +
P+
m
2L
ss

m
[ +X
1
(k) (11)

X
2
(k +1) = t j-
Po
rm
2
X
1
(k) -

s
+R
L
L
ss
X
2
(k)[ +X
2
(k)
(12)

where t is sampling time and is chosen as 0.0001sec for
simulation purposes. The system matrices (A
i
), input
matrices (b
i
), output matrices (C
i
) for the given fuzzy
model is based on maximum and minimum values (i=1,
N. SELVAGANESAN, R. SARASWATHY RAMYA / ELEKTRIKA, 11(1), 2009, 38-43
40
2) of the premise variables is developed and they are as
follows.

s L
1
ss
1
s L
1
ss
R +R
- -S
L
A =
R +R
S -
L






and
s L
2
ss
2
s L
2
ss
R +R
- -S
L
A =
R +R
S -
L







m 1
ss 1
S
L b =
0





and
m 2
ss 2
S
L b =
0






[ ]
1
C = 1 0
and
[ ]
2
C = 0 1

Fuzzy sets chosen for the model are
r 1
1
1 2
-S
F =
S -S
and
r 2
2
2 1
-S
F =
S -S

Which is obtained from the expression

S
1
F
1
+ S
2
F
2
=
r
(13)



Final output of the fuzzy model is computed using
parallel distributed compensation and is obtained as


2 1
1 2 2 1
F F
F Y F Y
YY
+
+
=
(14)
From the fuzzy modeling
r
qs
i and
r
ds
i are determined.
Based on
r
qs
i and
r
ds
i values u
qs
and u
ds
are calculated. By
Parks transformation, the 2-axis quantity is converted in
3-axis quantity. Hence, three phase currents and voltages
are obtained.

4. SIMULATION RESULTS
The specification of the PMSG drive is given in
Appendix [I]. The fuzzy modeling for PMSG is simulated
using MATLAB. The q-axis current, d-axis current, q-
axis voltage, d-axis voltage, three phase currents, three
phase voltages are obtained and compared with
mathematical model simulation results.

4.1 Without load disturbance
The mathematical model and the fuzzy model are
evaluated by superimposing their responses in the same
graph for the load resistance of 100. The i
qs
, zoomed
view of
r
qs
i ,
r
ds
i , u
qs
, zoomed view of u
qs
, u
ds,
zoomed
view of u
ds
, three phase currents and three phase voltages
are shown in the Figure 1 to Figure 9 respectively. A load
of 100 is applied to mathematically modeled PMSG.
The results such as
r
qs
i = 0.425A,
r
ds
i = 0.035A, u
q
=
42.96V, u
d
=0.336V, three phase currents of magnitude
0.425 A, three phase voltages of magnitude 42.95V are
obtained. Similarly for fuzzy model, the results such as
r
qs
i = 0.425A,
r
ds
i = 0.033A, u
q
= 42.96V, u
d
=0.336V,
three phase currents of magnitude 0.425A and three phase
voltages of magnitude 42.92V are also obtained.

Figure 1. q-axis current

Figure 2. Zoomed view of q-axis current

Figure 3. d-axis current

Figure 4. q-axis voltage
Fuzzy model
Mathematical model
Mathematical model
Fuzzy model
N. SELVAGANESAN, R. SARASWATHY RAMYA / ELEKTRIKA, 11(1), 2009, 38-43
41

Figure 5. Zoomed view of q-axis current

Figure 6. d-axis voltage

Figure 7. Zoomed view of d-axis voltage

Figure 8. Three phase current


Figure 9. Three phase voltage
4.2 With load disturbance
The PMSG drive is subjected to load disturbance by
varying the load resistance from 100 to 50 at 1 sec
and 50 to 10 at 2sec. Figure 10 to Figure 15 shows the
r
qs
i ,
r
qs
i , u
qs
, u
ds
, three phase currents and three phase
voltages under changes in load. Initially a load of 100 is
applied to PMSG and the responses are observed. A
change in load of 50 and 10 is applied at 1 sec and 2
sec respectively and its corresponding mathematical
model and fuzzy model responses are observed.
In mathematical model, when the load is 100 ,
r
qs
i =
0.425A,
r
ds
i = 0.035A, u
q
= 42.96V, u
d
=0.336V,
three phase currents of magnitude 0.425A and three phase
voltages of magnitude 42.96V are obtained. With the
change in load to 50 at 1 sec
r
qs
i = 0.83A,
r
ds
i =
0.13A, u
q
= 42.835V, u
d
= 0.66V, three phase currents of
magnitude 0.8374 A and three phase voltages of
magnitude 42.83V are obtained. The change in load of 10
is introduced at 2 sec and its corresponding variations
in
r
qs
i

= 2.6A,
r
ds
i = 1.96A, u
q
= 41.18V, u
d
= 2.04V,
three phase currents of magnitude 3.108A and three phase
voltages of magnitude 41.24V are also obtained.
When the fuzzy modeled PMSG is subjected to 100
load,
r
qs
i = 0.425A,
r
ds
i = 0.035A, u
q
= 42.75V, u
d

=0.32V, three phase currents of magnitude 0.425A and
three phase voltages of magnitude 42.75V are obtained.
The change in load to 50 is introduced at 1sec and the
corresponding variations of
r
qs
i

= 0.825A,
r
ds
i = 0.144A,
u
q
= 42.825V, u
d
= 0.656V, three phase currents of
magnitude 0.8374A and three phase voltages of
magnitude 42.842V are obtained. With the change in load
to 10 at 2 sec
r
qs
i = 2.40A,
r
ds
i = 2A, u
q
= 41V, u
d
=
1.875V, three phase currents of magnitude 3.108A, three
phase voltages of magnitude 41.24V are obtained.
From the results it is observed that for a small
variation in
r
qs
i ,
r
ds
i , u
q
and u
d
,

the net three phase
currents and voltages are not affected much. Also it is
found that, when PMSG drive is subjected to load
disturbance the responses of fuzzy modeled PMSG tracks
the mathematical model responses accurately.

Fuzzy model
Mathematical model
Mathematical model
Fuzzy model
N. SELVAGANESAN, R. SARASWATHY RAMYA / ELEKTRIKA, 11(1), 2009, 38-43
42

Figure 10. q-axis (100 to 50 at 1sec and 50
to 10 2sec)


Figure 11. d-axis current (100 to 50 at 1sec and 50
to 10 at 2sec)

Figure 12. q- axis voltage (100 to 50 at 1sec and 50
to 10 at 2sec)

Figure 13. d-axis voltage (100 to 50 at 1sec and 50
to 10 at 2sec)





Figure 14. Three phase current (100 to 50 at 1sec and
50 to 10 at 2sec)

Figure 15. Three phase voltage (100 to 50 at 1sec and 50
to 10 at 2sec)

5. CONCLUSION

The design methodology of fuzzy based modeling for
PMSG drive is presented. The Takagi-Sugeno based
singleton fuzzy model is developed and simulated for
load disturbances and its responses are compared with
mathematical model responses. From the simulation it is
observed that both the responses are matching closely.
REFERENCES
[1] Sergey Edward Lyshevski, Electromechanical
Systems, Electric Machines, and Applied
Mechatronics, CRC Press, FL 2001.
[2] Kai Zhang, Hossein M. Kojabadi and Peter Z.Wang,
Liuchen Chang, "Modeling of a Converter-
Connected Six-Phase Permanent Magnet
Synchronous Generator", IEEE PEDS 2005, pp.
1096-1100.
[3] Aleksandr Nagorny and Narajan V. Dravid, High
speed Permanent magnet synchronous
motor/generator design for flywheel applications,
National research council associateship award,
Cleveland, Ohio, 2005.
[4] Sergey Edward Lyshevski and Alberico Menozzi,
Control of Permanent magnet brushless DC motor,
Proceedings of the American Control Conference,
Arlington, VA June 25-27, 2001,pp. 2150-2155.
[5] B. Dehkordi, A. M. Gole and T. L Maguire
Permanent Magnet Synchronous Machine Model
for Real- Time Simulation, International
Conference on Power Systems Transients (IPST05),
Montreal, Canada, June 19-23, 2005
Mathematical model
Fuzzy model
Fuzzy model
Mathematical model
Mathematical model
Fuzzy model
N. SELVAGANESAN, R. SARASWATHY RAMYA / ELEKTRIKA, 11(1), 2009, 38-43
43
[6] Kazuo Tanaka and Hua O. Wang, Fuzzy Control
System Design and Analysis: A linear Matrix
Inequality Approach, John Wiley & sons, 2001
[7] P. C. Sen, "Electric Motor Drives and Control Past,
Present, and Future", IEEE Transactions on
Industrial Electronics, vo1.37, No.6, pp.562-575,
1990.
[8] R. Krishnan, "Selection Criteria for Servo Motor
Drives", IEEE Transactions on Industry
Applications, Vo1.M-23, No.2, pp. 270-275, 1987.
[9] G. R. Slemon, Electric Machines and Drives,
Addison-Wesley Publication Company, pp. 503511,
1992.
[10] L. A. Zadeh, Outline of a new approach to the
analysis of complex systems and decision processes,
IEEE Trans. SMC, SMC3, pp. 28-44, 1973.
[11] M. Sugeno and T. Yasukawa, A fuzzy-logic-based
approach to qualitative modeling. IEEE Trans. Fuzzy
Systems, FS-1, pp. 7 - 3 1, , l993.
[12] Zhu Wenbiao and Sun Zengqi, Fuzzy modeling for
complex process, IEEE Proc. of the 3
rd
World
Congress on Intelligent Control and Automation, pp
1271-2175, 2000.
[13] T. Takagi and M. Sugeno, Fuzzy Identification of
Systems and Its Applications to Modeling and
Control, IEEE SMC, vol.15, no.1, pp.116-132,
1985.
[14] Kuang-Yow Lian, Chian-Song Chiu, Tung-Sheng
Chiang, and Peter Liu Secure Communications of
Chaotic Systems with Robust Performance via Fuzzy
Observer-Based Design, IEEE Transactions on
fuzzy systems, Vo1.9, No.1, pp. 212-221, 2001.
[15] V. Natarajan, P. Kanagasabapathy, N. Selvaganesan
and P. Natarajan, Fuzzy observer design with n-shift
multiple key for cryptography based on 3-D hyper
chaotic oscillator, Iranian Journal of Fuzzy Systems,
Vol. 3, No. 2 pp. 21-32, 2006.
[16] Zhang Bo, Li Zhong and Mao Zong Yuan, A type
of fuzzy modeling of the chaotic system of
Permanent Magnet Synchronous Motor IEEE Conf.,
pp 880-883.
[17] Jacek Kabzifski, Fuzzy modeling of disturbance
torques/forces in rotational/linear interior permanent
magnet synchronous motors, Proc. On EPE-2005,
pp 1-10.









Appendix I
PMSG Specification

No. of phases 3
No. of poles 4
Rated Voltage (V) 43 V
Rated Current 6.9 A
Rated Power 135 W
Stator Resistance (r
s
) 0.5
Self Inductance (L
ss
) 0.01 H
Flux Linkage (
m
) 0.069 N-m-A
-1

Viscous Friction co-
efficient (B
m
)
0.0006 N-m-s-rad
-1

Moment of Inertia (J)
0.0075 Kg-m
2

Rated Speed () 399 rad-s
-1
Model H234-G Kollmorgen