See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224584740

Comparative Study of Fuzzy Logic Speed Controller in Vector Controlled

PMSM Drive: Minimum Number of Fuzzy Rule-Base

Conference Paper · August 2009

DOI: 10.1109/CITISIA.2009.5224229 · Source: IEEE Xplore

CITATIONS READS

16 315

3 authors:

Siti Noormiza Binti Mat Isa Zulkifilie Ibrahim

Technical University of Malaysia Malacca Technical University of Malaysia Malacca
19 PUBLICATIONS 146 CITATIONS 68 PUBLICATIONS 1,020 CITATIONS

SEE PROFILE SEE PROFILE

Fazlli Patkar
Technical University of Malaysia Malacca
16 PUBLICATIONS 129 CITATIONS

SEE PROFILE

All content following this page was uploaded by Siti Noormiza Binti Mat Isa on 16 October 2015.

The user has requested enhancement of the downloaded file.

 2009 Conference on Innovative Technologies in Intelligent Systems and Industrial Applications (CITISIA 2009)
Monash University, Sunway campus, Malaysia, 25th & 26th July 2009.

Comparative Study of Fuzzy Logic Speed Controller in Vector

Controlled PMSM Drive: Minimum Number of Fuzzy Rule-Base
Siti Noormiza Mat Isa, Zulkifilie Ibrahim, Fazlli Patkar

Abstract— This paper presents a comparative study on fuzzy on expert knowledge into an automatic control strategy.
rule-base of fuzzy logic speed control with vector-controlled There is no need of mathematical models to deal with a
Permanent Magnet Synchronous Motor (PMSM) drive. Fuzzy problem, but skill is needed to create the rules in a particular
rule-base design is viewed as control strategy. All fuzzy rules FL controller [1], [6], [9]. Reference [13] also mentioned
contribute to some degree in obtaining the desired
that FL controller essentially is a multi-parameter controller,
performance. However, some rules fired weakly do not
contribute significantly to the final result and can be whose performance depends on the selected shape of
eliminated. ‘Standard design’ of fuzzy rule-base of fuzzy logic membership function, rule base and scaling factor. This
controller is identified from numerous publications which is 49 means a comparison and verification on the varying
rules is used in order to obtain great speed tracking response. It parameters can be described in order to determine the
is possible to minimize the complexity of controller’s design by strength of the controller itself. Hence, it can be concluded
reducing number of rule-base from standard 49 rules to 9 rules.
that the varying parameters in terms of shape of membership
Simulation results that verify appropriateness of the approach
are included. A comparison of speed response between both function, number of rule-base and scaling factor
designs is studied. Sensitivity of the designs to load disturbance determination have great influence in obtaining the strength
and changes in command settings is studied and also being of FL controller with intention to attain the desired
compared to demonstrate the effectiveness of the reduced rule- performance. This point is also supported by [14] which
base. mentioned that all fuzzy rules contribute to some degree to
the final inference or decision, however, some rules fired
I. INTRODUCTION weakly do not contribute significantly to the final decision
Nowadays, fuzzy logic (FL) speed controller is well- and may be eliminated (reduced). Therefore, a study of
known with its superior performance in the research minimizing the fuzzy rule-base is presented in this paper.
community worldwide. It has been proven by many
simulations and experimental verifications by numerous The performance of vector controlled three-phase current-
publications with diversity of industrial drive applications fed induction motor, from the control point of view, can be
such as high performance DC drives [1], vector controlled equivalent to the dc motor drive whereby it decouples flux
Induction Motor [2]-[4], Permanent Magnet Synchronous and torque control by instantaneous stator current space
Motor [5]-[9], Brushless DC Motor [10] and Switched vector components can be achieved relatively easily [3], [8],
Reluctance Motor [11], [12]. [15]. Therefore, beginning from the successful performance
of vector control technique to induction, it may be
Various publications present comparison between the considered that this kind of speed control loop design can be
operation of drive with speed control by FL and PI implemented to any type of AC drives including Permanent
techniques. Then the main conclusion verified that FL Magnet Synchronous Motor [6]-[9], [16] and Switched
control determines superior performance [2]-[4], [6], [9]. Reluctance Motor [12].
However, the papers are focused just on the speed
performance meanwhile the parameters of the FL controller The drive used in the study comprises a permanent
are distinctive and fix. magnet synchronous motor with vector control technique.
Current control of PWM Voltage Source Inverter (VSI) is
Theoretically, FL is based on human reasoning, providing performed in stationary reference frame, using hysteresis
algorithms which can convert a set of linguistic rules based current controllers.

From the literature that have been reviewed, most papers

Manuscript received June 26, 2009.
Siti Noormiza Mat Isa is with the Electrical Engineering Department, utilize triangular membership functions with overlap, seven
Universiti Teknikal Malaysia Melaka, Malaysia (e-mail: for speed error and seven for change in speed error, so that
[email protected]). 7x7 = 49 rules are produced [5], [6], [16], [17]. This means
Zulkifilie Ibrahim is with the Electrical Engineering Department,
Universiti Teknikal Malaysia Melaka, Malaysia (phone: 606-555 2202; fax:
49 rules is a standard approach for the FL speed control with
606-555 2222; e-mail: [email protected]). PMSM drive application.
Fazlli Patkar is with the Electrical Engineering Department, Universiti
Teknikal Malaysia Melaka, Malaysia (phone: 606-555 2352; e-mail: Based on the above points, this paper aims to attempt to
[email protected]).
provide a comparison regarding speed drive operation with

978-1-4244-2887-8/09/$25.00 ©2009 IEEE 112

 2009 Conference on Innovative Technologies in Intelligent Systems and Industrial Applications (CITISIA 2009)
Monash University, Sunway campus, Malaysia, 25th & 26th July 2009.

utilizing 49 rules as ‘standard design’ and 9 rules as ‘case • Rated step speed command is applied at zero time under
design’ in FL speed controller. All simulation work is no-load conditions;
realized in MATLAB program. • Rated load torque is applied at time instant t = 0.04 sec;
• Speed reference is reduced by 10% of the previous
II. DESCRIPTION OF THE SYSTEM setting at time instant t = 0.08 sec;
Fig. 1 presents a speed control system of the vector • Initial step speed command is changed to 209 rad/s, 104
controlled PMSM drive. The rotor speed, ωr is compared rad/s and 52 rad/s speed command.
with ωr* and the resulting error is processed in the
In addition, the reverse operation is also tested.
controller. The output of controller is reference torque, T*
which is then has been limited by a limiter in order to
generate the q-axis reference current, iqs*. (Refer to Fig.
2(a)). At the mean time, d-axis reference current, ids* is set
to zero. Both d-axis and q-axis stator currents generate three
phase reference currents (ia*, ib* and ic*) through Park’s
Transformation which are compared with sensed winding
currents (ia, ib and ic) of the PMSM. The current errors are
fed to hysteresis current controllers which generate
switching signals for the voltage source inverter. Thus, by (a)
obtaining winding currents of the system, the speed response
is obtained.

(b)

Fig. 2. (a) Fuzzy Logic Controller (b) Internal Structure of FLC

Fig. 1. Configuration of FLC based vector controlled PMSM drive

III. DESIGN OF FUZZY LOGIC SPEED CONTROLLER
This paper uses the standard approach of FL controller.
The main goal of the control system is to track the
The width of triangular membership function is divided
command speed by reducing the complexity of fuzzy rule-
equally in a range with overlap, either 7x7 rules or 3x3 rules.
base design of fuzzy logic controller while maintaining the
Input and output scaling factors are determined through
performance obtained by ‘standard design’.
initial design applied to rated speed until the speed response
matches as closely as possible to the step rated speed A. Scaling Factor Calculation
command [4], [7] by implementing 49 rules. The role of scaling factor is similar to gain coefficients in
a conventional controller, and affects the stability,
The standard structure of fuzzy logic is given in Fig. 2 oscillations and damping of the system, hence needs to be
with inputs are speed error, e and change in speed error, ce chosen with utmost care [17]. Three scaling factors
and output is change in q-axis reference current, ∆iqs* [2], represented by Ge, Gce and Gcu are chosen using known
[4]-[6], [11], [16]. The FLC executes the rule base taking the motor data given for fuzzification, as well as for obtaining
fuzzy variables e and ce as the inputs and quantity of ∆iqs* as the actual output of the command current. There is always an
the output are processed in the defuzzification unit as input limitation for FLC, so that for convenience inputs and
illustrated in Fig. 2(b). For the present work, Mamdani type output are always normalized into [-2,2] interval. Rated
fuzzy inference is used [4]. speed of the motor is 314 rad/s and an assumption is made
that this value is the maximum speed of operation of the
A comparison of FL speed controller for different number motor. Thus, maximum speed error is 314 for start-up from
of fuzzy rule-base which is ‘standard design’ and ‘case standstill and the scaling factor for the speed error is
design’ is studied. The focus is on the speed response during obtained as [4], [7]:
start-up under no load, load disturbance operation and step
change in reference speed setting. The investigation of fuzzy Ge = 1/314 ≈ 0.00318 (1)
controllers is also carried out for several selected speed
command in such a way to further strengthen the result Scaling factor for the change in speed error is calculated
obtained. The research details are as follows:

113
 2009 Conference on Innovative Technologies in Intelligent Systems and Industrial Applications (CITISIA 2009)
Monash University, Sunway campus, Malaysia, 25th & 26th July 2009.

on the basis of rated inertia and maximum torque that the IF ‘speed error is NS’ AND ‘change in speed error is PS’
motor is allowed to develop, taking sampling time 20μs. THEN ‘change in q-axis reference current is ZE’.

Te max = Jn/P(∆ω/Ts) →∆ω = 2.3438 rad/s TABLE II

MATRIX OF ‘CASE DESIGN’
Gce = 1/ce = 1/(e(Ts) – e(0)) = 1/∆ω = 0.4267 (2)
Speed error, e
Output scaling factor is set to Gcu = 1. N ZE P
N N N ZE

Change in

error, ce
speed
ZE N ZE P
B. Design of Fuzzy Rule-Base
P ZE P P
In this study, in order to determine the controller output
from the measured system variables, a fuzzy relation matrix
is established. The fuzzy relation matrix gives a relationship
TABLE III
between fuzzy set characterizing controller inputs and fuzzy PMSM TEST MOTOR
set characterizing controller outputs. Below, the fuzzy rule
base matrix for ‘standard design’ and ‘case design’ are Parameter Value
represented by Table I and Table II respectively. As Maximum torque 10 Nm
mentioned before, the rules for the ‘standard design’ are Rated torque 1.7 Nm
determined through literature review from many Rated current 2.7 A
publications. Meanwhile, the rules for the ‘case design’ Maximum current 10 A
parameters are determined by standard approach with Rated speed 314 rad/s (3000 rpm)
reducing the number of fuzzy rule-base. The linguistic
Inertia 0.000256 kgm2
elements used are the same as those used in most of the
publications [2], [9]. Fixed-step mode is chosen for the Winding resistance 2.67 Ω
computational time interval. Numerical method for solving Winding inductance 11.5 mH
differential equations is Dormand-Prince [8] and simulating Magnet flux 0.1210 Vs/rad
period is 0.1 second. Nominal parameters for the test motor Rated frequency 150 Hz
are showed in Table III. Pole pairs 3
DC link voltage 320 V
TABLE I
MATRIX OF ‘STANDARD DESIGN’

Speed error, e
NL NM NS ZE PS PM PL

NL NL NL NL NL NM NS ZE
Change in speed error, ce

NM NL NL NL NM NS ZE PS
NS NL NL NM NS ZE PS PM
ZE NL NM NS ZE PS PM PL
(a)
PS NM NS ZE PS PM PL PL
PM NS ZE PS PM PL PL PL
PL ZE PS PM PL PL PL PL

Seven terms are assigned in Table I: NL, negative large;

NM, negative medium; NS, negative small; ZE, zero; PS,
positive small; PM, positive medium; and PL, positive large.
Three terms are assigned in Table II: N, negative; ZE, zero;
and P, positive. Each fuzzy variable is a member of the
subsets with a degree of membership µ varying between 0
(b)
and 1. As mentioned before, for convenience, the rules have
been written in matrix form and should be interpreted as Fig. 3. Membership functions for fuzzy sets of speed error, change in speed
error and change in q-axis reference current: (a) ‘standard design’, (b) ‘case
(Refer to Table I ): design’

114
 2009 Conference on Innovative Technologies in Intelligent Systems and Industrial Applications (CITISIA 2009)
Monash University, Sunway campus, Malaysia, 25th & 26th July 2009.

The shape of the fuzzy sets on the two extreme ends of the
universe of discourse is taken as trapezoidal whereas all
other intermediate fuzzy sets are triangular with overlap to
each other as standard approach and the constant are divided
equally in the range of -2 and 2 based on the number of
membership functions either seven or three as illustrated in
Fig. 3 [1], [3]. This operation can be performed by several
methods which is max-min reasoning and center of gravity
(or centroid) defuzzification method are used, as those
methods are most frequently used in many literatures [1],
[4], [6].

All the scaling factors, shape of membership function,

method of fuzzification and method of defuzzification are
predefined and kept constant during the research except the
number of rules.

(b)
IV. SIMULATION RESULTS
The study of vector-controlled PMSM drive with
implementing ‘standard design’ and ‘case design’ of FL
speed controller is carried out in the MATLAB (Simulink
and Fuzzy Toolbox) program. All study based on rated value
of the test motor model as shown in Table III. The
investigations are aimed at studying the speed responses of
the fuzzy speed controller with implementation of ‘standard
design’ and ‘case design’.

The response to step changes in speed command,

subsequent rated load torque application and step reduction
of speed command to 0.9 times the previous value are
studied next for four arbitrarily selected speed command:
rated speed (314 rad/s), two third of rated speed (209 rad/s),
one third of rated speed (104 rad/s) and one sixth of rated
speed (52 rad/s). Comparison of the drive behaviour under
‘standard design’ and ‘case design’ speed control are
(c)
performed by overlapping and zooming speed responses of
the types illustrated in Figs. 4, 5 and 6.

(d)

Fig. 4. Comparison of speed response obtained during start-up by ‘standard

(a) design’ and ‘case design’ for four different speeds: (a) at rated, 314 rad/s (b)
at 209 rad/s (c) at 104 rad/s (d) at 52 rad/s

115
 2009 Conference on Innovative Technologies in Intelligent Systems and Industrial Applications (CITISIA 2009)
Monash University, Sunway campus, Malaysia, 25th & 26th July 2009.

Fig. 4 presents good speed responses of the motor drive

during start-up from standstill and accelerated to the speed
command without load. The results show almost similar
performances are obtained by both designs and the responses
are clearly consistent with no overshoot and negligible
steady state error. It is evident that the ‘case design’ has
acceptable performance and it seems settled at the same
moment whereby the difference of settling time between
both controllers is approximately 5 ms.

After the responses settling down at the speed command,

the system is loaded with rated load torque which is 1.7 Nm
at instant time, t = 0.04 sec and the results are showed in Fig.
5. Both controllers reject load disturbance rapidly without
overshoot and almost zero steady state error. A dip of
approximately 5 rad/s is formed. The dip of ‘standard
(b)
design’ is slightly deeper than ‘case design’ but it recovers
slightly faster than ‘case design’ about 5 ms. It can be
clearly seen that the ‘case design’ produces better
undershoot response compared to ‘standard design’ during
load disturbance but no longer recovery time. It is also
shows that both designs of controllers produce almost
consistent undershoot for different step speed command
during load disturbance.

(c)

(a) (d)

Fig. 5. Comparison of speed response to step rated load torque application

obtained by ‘standard design’ and ‘case design’ for four different speeds:
(a) at rated, 314 rad/s (b) at 209 rad/s (c) at 104 rad/s (d) at 52 rad/s

116
2009 Conference on Innovative Technologies in Intelligent Systems and Industrial Applications (CITISIA 2009)
Monash University, Sunway campus, Malaysia, 25th & 26th July 2009.

(d)
(a)
Fig. 6. Comparison of speed responses of 10% reduction to speed
command obtained by ‘standard design’ and ‘case design’ for four different
speeds: (a) at rated, 314 rad/s (b) at 209 rad/s (c) at 104 rad/s (d) at 52 rad/s

The motor operates in steady state condition as Fig. 5.

Then, a step speed command reduction, equal to 10% of the
previous reference setting is applied at t = 0.08 sec. The
results are given in Fig. 6 which included with tests for four
different speed commands. It verifies that both controllers
yield identical speed response for zero undershoot and small
steady state error. It is also clear that the responses are
consistent with settling time approximately 0.095 sec for
four different speed commands.

Next, reverse operation is tested to the PMSM drive. Fig.

7 shows the forward operation at start-up and reverse
(b) operation at 0.05 sec while Fig. 8 shows the other ways. It
obviously shows that both designs perform quite similar
performances. There are no overshoot and no undershoot,
the settling time of both designs are relatively similar and
negligible steady state error.

(c)

Fig. 7. Comparison of speed responses during forward and reverse

operation obtained by ‘standard design’ and ‘case design’ at rated speed
command

117
 2009 Conference on Innovative Technologies in Intelligent Systems and Industrial Applications (CITISIA 2009)
Monash University, Sunway campus, Malaysia, 25th & 26th July 2009.

[4] M. N. Uddin, T. S. Radwan, M. A. Rahman, “Performances of Fuzzy

Logic Based Indirect Vector Control for Induction Motor Drive,”
IEEE Trans. On Industry Applications, vol. 38, pp. 1219-1225,
Sept./Oct. 2002.
[5] A. Karakaya, E. Karakas, “Performance Analysis of PM Synchronous
Motors Using Fuzzy Logic and Self Tuning Fuzzy PI Speed
Controls,” The Arabian Journal for Science and Engineering, vol. 33,
pp. 153-177, Apr. 2008.
[6] A. G. Aissaoui, M. Abid, H. Abid, A. Tahour, A. K. Zeblah, “A Fuzzy
Logic Controller for Synchronous Machine,” Journal of Electrical
Engineering, vol. 58, pp. 285-290, 2007.
[7] M. N. Uddin, T. S. Radwan, M. A. Rahman, G. H. George, “Fuzzy
Logic Based Position Control of Permanent Magnet Synchronous
Motor,” in Canadian Conference on Electrical & Computer
Engineering, vol. 1, pp. 93-97, 2000.
[8] T. Pham-Dinh, T. Nguyen-Thanh, “Fuzzy Speed Controller for Rotor
Flux Oriented Control of Permanent Magnet Synchronous Machine,”
in Proc. ISEE 2007, pp. 72-78, Oct 2007.
[9] Ì. Güney, Y. Oğuz, F. Serteller, “Dynamic Behaviour Model of
Fig. 8. Comparison of speed responses during reverse and forward Permanent Magnet synchronous Motor Fed by PWM Inverter and
operation obtained by ‘standard design’ and ‘case design’ at rated speed Fuzzy Logic Controller for Stator Phase Current, Flux and Torque
command Control of PMSM,” in Proc. IEMDC’01, pp. 479-485, 2001.
[10] R. N. Tuncay, Z. Erenay, M. Yilmaz, Ö.Ustun, “Rapid Control
Prototyping Approach to Fuzzy Speed Control of Brushless DC
Motor,” in Proc. ELECO’03, Dec 2003.
Based on above results obtained, it can be seen that the [11] M. G. Rodrigues, W. I. Suemitsu, P. Branco, J. A. Dente, L. G. B.
‘case design’ can provide acceptable performance as the Rolim, “Fuzzy Logic Control of a Switched Reluctance Motor,” in
‘standard design’ performance in the FL control process. It Proc. ISIE’97, vol. 2, pp. 527-531, Jul. 1997.
[12] Z. Yingchao, Z. Zhengming, G. Wei, S. Xiaoying, “High Performance
is essential to note that the difference of settling time of 5 ms
Position Control System Based on SR-PM Motor,” Tsinghua Science
during start-up and load disturbance do not give any and Technology, vol. 12, pp. 614-619, Oct. 2007.
significant changes in real-time environment. [13] Z. Ibrahim, E.Levi, D. Williams, “A Self-Tuning Method for Fuzzy
Logic Speed Controller in High Performance AC Drives,” in Proc.
UPEC’98, vol. 2, pp. 819-822, Sep. 1998.
[14] C. Hung, B. Fernández, “Minimizing Rules of Fuzzy Logic System by
V. CONCLUSIONS Using a Systematic Approach,” IEEE Std., 1993.
A fuzzy rule-base design of fuzzy logic speed control has [15] H. N. Nounou, H. Rehman, “Application of Adaptive Fuzzy Control
to AC Machines,” Applied Soft Computing, vol. 7, pp. 899-907, 2007
been studied for speed control of vector controlled PMSM [Online]. The sciencedirect website. Available:
drive. A set of fuzzy decision rules are formulated based on http://www.elsevier.com/locate/apenergy
the literature review of the controller’s design. In this paper, [16] B. Singh, B. P. Singh, S. Dwivedi, “DSP Based Implementation of
Fuzzy Precompensated PI Speed Controller for Vector Controlled
two types of fuzzy speed controller named ‘standard design’ PMSM Drive,” IEEE Industrial Electronics and Applications, pp. 1-6,
and ‘case design’ is studied and being compared based on May 2006.
speed response during start-up under no load, load [17] A. Karakaya, E. Karakas, “The Speed Control of Permanent Magnet
Synchronous Motor Using Fuzzy Logic and Self Tuning Fuzzy PI
disturbance and changes in command settings. The Controller,” in Proc. ELECO’05, Dec. 2005.
simulation study is realized in MATLAB environment. The [18] Unit-V Applications of Fuzzy Logic Control. [Online]. Available:
investigation of fuzzy controllers is carried out based on gapp.files.wordpress.com/2008/05/unit-5.pdf
several selected speed response. The results are plotted in
the same speed graph with intention to make details
comparison based on visual observation on settling time,
undershoot and recovery time. It is showed that both types of
controller produce similar performance, thus, it is feasible to
minimize the complexity of fuzzy logic controller by
reducing the number of fuzzy rule-base from 49 rules to 9
rules.

REFERENCES
[1] J. N. Lygouras, P. N. Botsaris, J. Vourvoulakis, V. Kodogiannis,
“Fuzzy Logic Controller Implementation for a Solar Air-conditioning
System,” Applied Energy 84, vol. 20, pp. 1305-1318, 2007 [Online].
The sciencedirect website. Available:
http://www.elsevier.com/locate/apenergy
[2] V. Kumar, R. R. Joshi, “Hybrid Controller Based Intelligent Speed
Control of Induction Motor,” Journal of Theoretical and Applied
Information Technology, 2005, pp. 71-75. [Online]. Available:
http://www.jatit.org
[3] H. A. F. Mohamed, W. P. Hew, “A Fuzzy Logic Vector Control of
Induction Motor,” in Proc. TECON 2000, vol. 3, pp. 324-328.

118

View publication stats