Improved Rotor Speed Brushless DC Motor Using Fuzzy Controller

Indonesian Journal of Electrical Engineering and Informatics (IJEEI)
Vol. 3, No. 2, June 2015, pp. 78~88
ISSN: 2089-3272  78
Received January 12, 2015; Revised March 30, 2015; Accepted April 23, 2015
Improved Rotor Speed Brushless DC Motor Using
Fuzzy Controller
Jafar Mostafapour1
*, Jafar Reshadat1,3
, Murtaza Farsadi2
1
Azerbaijan Regional Electric company, Tariz, Iran
2
Department of Electrical Engineering, Urmia University
3
Department Management, Science and Research Branch Islamic Azad University, West Azerbaijan, Iran
e-mail: j.mostafapour.a@gmail.com
Abstract
A brushless DC (BLDC) Motors have advantages over brushed, Direct current (DC) Motors and,
Induction motor (IM). They have better speed verses torque characteristics, high dynamic response, high
efficiency, long operating life, noiseless operation, higher speed ranges, and rugged construction. Also,
torque delivered to motor size is higher, making it useful in application where space and weight are critical
factors. With these advantages BLDC motors find wide spread application in automotive appliance,
aerospace medical, and instrumentation and automation industries This paper can be seen as fuzzy
controllers compared to PI control BLDC motor rotor speed has improved significantly and beter result can
be achieve
Keywords: BLDC, matlab/simulink, PID controller, PID fuzzy controller
1. Introduction
Brushless motor technology makes it possible to achieve high reliability with high
efficiency, and for a lower cost in comparison with brush motors. Although the brushless
characteristic can be apply to several kinds of motors AC synchronous motors, stepper motors,
switched reluctance motors, AC induction motors the BLDC motor is conventionally defined as
a permanent magnet synchronous motor with a trapezoidal Back, Electric Magnetic Fields
(EMF) waveform shape. Permanent magnet synchronous machines with trapezoidal Back-EMF
and (120 electrical degrees wide) rectangular stator currents are widely used as they offer the
following advantages first, assuming the motor has pure trapezoidal Back EMF and that the
stator phases commutation process is accurate, the mechanical torque developed by the motor
is constant secondly, the Brushless DC drives show a very high mechanical power density.
Brushless Direct Current (BLDC) motors are one of the motor types rapidly gaining popularity.
BLDC motors are used in industries such as Appliances, Automotive, Aerospace, Consumer,
Medical, Industrial Automation Equipment and instrumentation [1, 2], As the name implies,
BLDC motors do not use brushes for commutation; instead, they are electronically commutated.
BLDC motors have many advantages over brushed DC motors and induction motors
BLDC Motors are available in many different power ratings, from very small motors as
used in hard disk drives to larger motors used in electric vehicles. Purpose of this article IS
improve the performance of BLDC rotor speed By using fuzzy control and compared with PI
controller.
2. Working of BLDC Motor
The BLDC motor is an AC synchronous motor with permanent magnets on the rotor
(moving part) and windings on the stator (6 part). Permanent magnets create the rotor flux and
the energized stator windings create electromagnet poles. The rotor (equivalent to a bar
magnet) is attracted by the energized stator phase. By using the appropriate sequence to
supply the stator phases, a rotating field on the stator is created and maintained. This action of
the rotor - chasing after the electromagnet poles on the stator is the fundamental action used in
synchronous permanent magnet motors [2]. The lead between the rotor and the rotating field
must be controlled to produce torque and this synchronization implies knowledge of the rotor
position.

IJEEI ISSN: 2089-3272 
Improved Rotor Speed Brushless DC Motor Using Fuzzy Controller (Jafar Mostafapour)
79
Figure 1. A three-phase synchronous motor with a one permanent magnet pair pole rotor
On the stator side, stator is three phase similar to induction motor These offer a good
compromise between precise control and the number of power electronic devices required to
control the stator currents. For the rotor, a greater number of poles usually create a greater
torque for the same level of current. On the other hand, by adding more magnets, a point is
reached where, because of the space needed between magnets, the torque no longer
increases. The manufacturing cost also increases with the number of poles. As a consequence,
the number of poles is a compromise between cost, torque and volume [3]. Permanent magnet
synchronous motors can be classified in many ways, one of these that is of particular interest to
us is that depending on back EMF profiles: Brushless Direct Current Motor (BLDC) and
Permanent Magnet Synchronous Motor (PMSM). This terminology defines the shape of the
back EMF of the synchronous motor. Both BLDC and PMSM motors have permanent magnets
on the rotor but differ in the flux distributions and back EMF profiles.
To get the best performance out of the synchronous motor, it is important to identify the
type of motor in order to apply the most appropriate type of control is described. We have seen
that the principle of the BLDC motor is, at all times, to energize the phase pair which can
produce the highest torque. To optimize this effect the Back EMF shape is trapezoidal. The
combination of a DC current with a trapezoidal Back EMF makes it theoretically possible to
produce a constant torque. In practice, the current cannot be established instantaneously in a
motor phase; as a consequence the torque ripple is present at each 60°
degree phase
commutation [4, 5].
Figure 2. Electrical Waveforms in the Two Phase ON Operation and Torque Ripple
2.1. Mathematical Model of BLDCM
As shown in Figure 3, a dynamic equivalent circuit of the BLDC motor. For this model,
the stator phase voltage equations in the stator reference frame of the BLDC Motor are given as
in Equation (1, 5) which are provided below. The following assumptions are made:1) the three
phase windings are symmetrical, 2) magnetic saturation is neglected, 3) hysteresis and eddy
current losses is not considered, and 4) the inherent resistance of each of the motor windings is
R ,the self-inductance is L, and the mutual inductance is M.

 ISSN: 2089-3272
IJEEI Vol. 3, No. 2, June 2015 : 78 – 88
80
Figure 3. Dynamic equalent circuit













































































)(
)(
)(
00
00
00
)(
1




c
b
a
mr
c
b
a
C
b
a
c
b
a
c
b
a
f
f
f
i
i
i
R
R
R
U
U
U
ML
i
i
i
dt
d (1)
Where, Ua, Ub and Uc are the phase voltage of three-phase windings, ia, ib and ic are the
phase current, and ea, eb and ec are the back EMF.




3601300)2)(
6
(
3001801
1801201)()
6
(
18001)(




rr
r
rr
rraf






(2)
Electrical power of motor can be calculated using Equation (3).
ccbbaa ieieieP  (3)
Electromagnetic torque can also be expressed as Equation (4). Speed is derived from
rotor position Өr as in Equation (5).
1)
2
()
2
( T
p
B
dt
d
p
jT rmre   (4)
)1)
2
()(
2
( T
p
BT
j
p
dt
d
rmer   (5)
rr
dt
d
  (6)
From the above equations, BLDC motor can be modeled [6].

IJEEI ISSN: 2089-3272 
81
3. A Review on Utilized Systems
3.1. Implementation of PID fuzzy Controller for BLDC
In this section implementation of Fuzzy Inference System for nonlinear fuzzy PID
control is explicated using control system toolbox of Simulink.
As mentioned before a fuzzy inference system maps known inputs to outputs using
fuzzy logic. For instance, mapping of a controller can be stated by a three dimensional diagram.
This diagram is called control surface. The following figure illustrates a hypothetical control
surface.
Figure 4. An example of a control surface
Error signal e(k) and error variation signal (e(k)-e(k-1)) are common inputs of FIS. The
output of FIS is a control operation which is inferred from fuzzy rules.
In our study, utilized system is a BLDC model, single input- single output, which is
discretized. The control objective is tracking reference signal.
3.2. Structure of Fuzzy PID Controller
The exploited fuzzy controller is a feedback loop which operates similar to PID which is
calculated by fuzzy inference. The closed loop structure in SimuLink is as follows. It can be
observed by typing the undergoing instruction.
Open-system ('Fuzzy_PID')
Figure 5. Closed loop structure in SimuLink
Three controllers depicted in the above figure are respectively conventional PID, linear
fuzzy PID controller and nonlinear fuzzy PID controller. We will see that it is necessary to design
conventional and linear fuzzy PID controllers to design nonlinear fuzzy controller.

 ISSN: 2089-3272
IJEEI Vol. 3, No. 2, June 2015 : 78 – 88
82
Parallel structure is utilized to implement fuzzy controller. It is a combination of fuzzy PI
and fuzzy PD controllers. The structure of fuzzy controller is demonstrated in figure below.
Figure 6. Structure of fuzzy controller
The magnitude of -(y(k)-y(k-1)) is used instead of signal changes. It is done in this way
to avoid direct stimulation of derivative signal by step changes in input reference. Two gain
blocks, GCE and GCU, are employed in the feedforward path. These two blocks guarantee that
error signal e, is used proportional when the fuzzy PI controller is linear
3.3. Design Procedure for Fuzzy PID Controller
Design of fuzzy controller includes configuration of fuzzy inference system and
substitution of GE, GCU, GCE and GU scaling factors. Here, the following steps are taken for
controller design.
a) Designing conventional PI controller
b) Designing equivalent linear fuzzy PID controller
c) Adjusting fuzzy inference system to obtain nonlinear control surface (designing
nonlinear fuzzy PID controller)
d) Optimum adjustment of nonlinear fuzzy PID controller
3.3.1. The First Step: Designing Conventional PID Controller
To implement PID controller, the parallel structure, which is shown below, is exploited.
1
1
s
p i d
s
T z z
K K K
z T z

 

.
The mentioned controller is implemented as follows in the SimuLink.
Figure 7. PID structure in the Simulink environment
Similar to fuzzy PID controller the input signal for derivative operator is –y(k).
PID coefficients might be adjusted manually or using adjustment rules. The following
instructions might be utilized to adjust PID controller coefficients in control tool box.
% Designing Conventional PID
C0 = pid(1,1,1,'Ts',Ts,'IF','B','DF','B'); % define PID structure
C = pidtune (plant,C0); % design PID
[Kp Ki Kd] = piddata(C); % obtain PID gains

IJEEI ISSN: 2089-3272 
83
3.3.2. Second Step: Designing Equivalent Fuzzy PID Controller
With FIS configuration and selecting four scaling coefficients, a fuzzy controller is
derived whose performance is exactly the same as conventional PID.
First off, fuzzy system is configured. As a result a linear control surface is achieved from
E and CE as inputs to U as output. The structure of utilized inference system is summarized as
follows.
a) Mamdani inference system is employed.
b) Algebraic multiplication is used instead of AND.
c) The input range is considered to be [-10,10]
d) The fuzzy sets are triangular and they intersect their neighbors in 0.5 membership
value.
e) The output range is [-20, 20].
f) The outputs are single-valued determined by sum of peak positions of input sets.
g) The center of gravity method is used for deffuzification.
Values of input and output ranges and membership function parameters must be
assigned so that the relation between input and output of the system is equal to an identity
function. In the next section the coefficients of fuzzy PID controller are derived by assuming the
identity function for relation of fuzzy inference system.
The following instruction is used to build fuzzy inference system.
%Designing Linear Fuzzy Inference System
FIS2 = newfis('FIS2','mamdani','prod','probor','prod','sum');
And the fuzzy rules are also defined as follows:
If E is Negative and CE is Negative then u is -20
If E is Negative and CE is Zero then u is -10
If E is Negative and CE is Positive then u is 0
If E is zero and CE is Negative then u is -10
If E is Zero and CE is Zero then u is 0
If E is Zero and CE is Positive then u is 10
If E is Positive and CE is Negative then u is 0
If E is Positive and CE is Zero then u is 10
If E is Positive and CE is Positive then u is 20
Here we utilized fuzzy tool box instructions to create FIS; however, corresponding GUI
might be used as well.
The 3d surface is achieved as follows:
Figure 8. 3D diagram of control surface in fuzzy PID controller
The input and output membership functions are shown below.

 ISSN: 2089-3272
IJEEI Vol. 3, No. 2, June 2015 : 78 – 88
84
Figure 9. Input and output membership functions for linear fuzzy PID controller
In the next stage, four scaling factors are calculated using coefficients of conventional
PID controller. The input-output relation in fuzzy inference system is considered to be in the
form of identity function; therefore, the corresponding relations are as shown in the following
equations.
p
i
d
k GCU GCE GU GE
k GCU GE
k GU GCE
   
 
 
If the maximum input step is considered as 1 the maximum error value would be 1.
Since input range equals to [-10, 10], considering GE=10, GCE, GCU and GU are derived from
following equations.
 2
10
4
2
p p i d
i
i
d
GE
k sqrt k k k
GCSE GE
k
k
GCU
GE
k
GU
GCE

 
 



The above values are calculated in the corresponding m-file using the following
instructions. They are used in Simulink plant file together with controller.
3.3.3. Third Step: Designing Fuzzy PID Controller with Nonlinear Control Surface
First we make sure that fuzzy PID controller is properly designed. Afterwards, FIS
adjustments such as, type, functions, membership, fuzzy rules and so on are changed so that
desired nonlinear control surface is achieved. For this purpose Sugeno inference system is
utilized. Moreover, for each input merely two states, positive and negative, are considered which
reduces the number of rules to four.
The fuzzy rule set is defined as follows.
If E is Negative and CE is Negative then u is -20
If E is Negative and CE is Positive then u is 0
If E is Positive and CE is Negative then u is 0
If E is Positive and CE is Positive then u is 20
-10 -8 -6 -4 -2 0 2 4 6 8 10
0
0.5
1
Membership Functions for input E -Linear Fuzzy PID
-10 -8 -6 -4 -2 0 2 4 6 8 10
0
0.5
1
Membership Functions for input CE -Linear Fuzzy PID
-20 -15 -10 -5 0 5 10 15 20
0
0.5
1
Membership Functions for output u -Linear Fuzzy PID

IJEEI ISSN: 2089-3272 
85
The 3D diagram of nonlinear control surface is depicted in figure, As shown in
Figure 10
Figure 10. 3D diagram of control surface for nonlinear fuzzy PID controller
As can be seen we have a nonlinear control surface. According to above mentioned
control surface, it can be seen that the control surface has considerable gain in the vicinity of
center of E and CE plane. As a result when error is small it will decrease more rapidly. When the
error is large, the variations of controller are small. It limits control operation and avoids
probable saturation. The membership functions for inputs of fuzzy inference system are
demonstrated as shown in Figure 11.
Figure 11. Input and output membership functions associated with nonlinear fuzzy PID controller
Figure 12. The response of closed loop system with conventional PID, linear fuzzy and
nonlinear fuzzy controllers
-10 -8 -6 -4 -2 0 2 4 6 8 10
0
0.5
1
Membership Functions for input E -Nonlinear Fuzzy PID
-10 -8 -6 -4 -2 0 2 4 6 8 10
0
0.5
1
Membership Functions for input CE -Nonlinear Fuzzy PID

 ISSN: 2089-3272
IJEEI Vol. 3, No. 2, June 2015 : 78 – 88
86
Figure 13. Closed loop system response with conventional PID, linear fuzzy PID and nonlinear
fuzzy PID controllers
Figure 14. The magnified response of closed loop system with conventional PID, linear fuzzy
PID and nonlinear fuzzy PID controllers
As shown in Figure 13, response of system with mentioned controllers is depicted for
step input and at t=1s.
As expected the response of the system with conventional PID and linear fuzzy PID are
the same. The response of the system with nonlinear fuzzy PID controller is faster than two
others; nevertheless, it does not show any improvements regarding overshoot.
.
3.3.4. Fourth Step: Optimum Adjustment of Nonlinear Fuzzy PID Controller
In this section system response is modified by changing the parameters of input
membership functions. The following results are achieved by changing membership functions
(changing the parameter related to membership function from 6 to 2)
%input E
FIS1 = addvar (FIS1,'input','E',[-10 10]);
FIS1 = addmf (FIS1,'input',1,'Negative','gaussmf',[2 -10]);
FIS1 = addmf (FIS1,'input',1,'Positive','gaussmf',[2 10]);
%input CE
FIS1 = addvar (FIS1,'input','CE',[-10 10]);
FIS1 = addmf (FIS1,'input',2,'Negative','gaussmf',[2 -10]);
FIS1 = addmf (FIS1,'input',2,'Positive','gaussmf',[2 10]);
The magnified step response is shown in Figure 16.

IJEEI ISSN: 2089-3272 
87
Figure 15. Block Diagram of BLDC
Figure 16. design fuzzy PI controller
As can be seen the response is faster with nonlinear fuzzy PID controller and it has
smaller overshoot. Comparing Figure 13 and 14 it can be concluded that the system response is
significantly improved by changing the parameters.
4. Simulation and Results
To design fuzzy PI controller, the following structure is used. As we know the structure
of a conventional PI is as follows:
Figure 17. The structure of a conventional PI
Comparing the two structures shows that the two following relationships are established
between the coefficients of the controller:
ki GE GU
kp GR GU
 
 
In the above equations, the coefficients of conventional PI can be obtained by having
fuzzy PI controller. In practice, we need to obtain the coefficients of the fuzzy PI controller. For
this purpose, by choosing a value for one of the coefficients such as GU, other coefficients can
be achieved. The selected values for GU controller determine the degree of nonlinearity for
fuzzy controller Here the input range for fuzzy inference system is considered as [-10.10] and
for this reason according to the reference signal amplitude and the estimated value that is

 ISSN: 2089-3272
IJEEI Vol. 3, No. 2, June 2015 : 78 – 88
88
approximately equal to 300, the value of GU = 1000 is considered that seems to be a good
value, an inference system of Sugeno is used as the fuzzy controller. Only two modes of
Positive and Negative are considered for each input and Positive, Zero and Negative modes are
considered for each output and the total rules have reduced to 4 rules.
To configure the Fuzzy inference system we run the following commands.
If E is Negative and CE is Negative then u is Negative
If E is Negative and CE is Positive then u is Zero
If E is Positive and CE is Negative then u is Zero
If E is Positive and CE is Positive then u is Positive
With the definition of fuzzy inference system and using fuzzy PI controller, the rotor
speed’s results are obtained as follows:
To compare the fuzzy PI controller and conventional PI controller, a PI with coefficients
equal to fuzzy PI is used. Results from both controllers and the reference input are shown in the
following figure. Values considered for PI controller are:
46.35, 1.22ki kp 
In the following figure, the system behavior is given with enlargement in one of the
corners.
Figure 18. Rotor speed with fuzzy PI controller
As can be seen, the use of fuzzy PI controller significantly improves the system
response and the system could follow the reference signal with very good accuracy. By
changing the parameters of the membership functions related to the fuzzy inference system, the
obtained results can be improved.
References
[1] AR Millner. Mult hundred horsepower permanent magnet brushless disc motors. In Proc. IEEE Appl.
Power Electron. Conf. (APEC’94). 1994: 351–355.
[2] Kun Wei, Zhengli Lou, Zhongchao Zhang, Research on the Commutation Current Prediction Control in
Brushless DC Motor.
[3] N Hemati, MC Leu. A complete model characterization of brushless dc motors. IEEE Trans. Ind.
Applicat. 1992; 28: 172–180.
[4] P Pillay, R Krishnan. Modeling, Simulation and Analysis of a Permanent Magnet Brushless dc Motor
Drive. Conference Record of IEEE/IAS Meeting. 1987: 8.
[5] Sung-In Park, Tae-Sung Kim, Sung-Chan Ahn, Dong-Seok Hyun. An Improved Current Control
Method for Torque Improvement of High- Speed BLDC Motor.
[6] P Philip, D Meenakshy K. Modelling Of Brushless DC Motor Drive Using Sensored And Sensorless
Control. IEEE Trans. J. Industry Application. 2012; 2(8).

Improved Rotor Speed Brushless DC Motor Using Fuzzy Controller

More Related Content

What's hot (19)

Similar to Improved Rotor Speed Brushless DC Motor Using Fuzzy Controller (20)

More from ijeei-iaes (20)

Recently uploaded (20)

Improved Rotor Speed Brushless DC Motor Using Fuzzy Controller