International Journal of Computer Applications (0975 – 8887)

Volume 109 – No. 10, January 2015

Brushless DC Motor Speed Control using both PI

Controller and Fuzzy PI Controller
Ahmed M. Ahmed Amr Ali-Eldin
MSc Student at Computers and Systems Engineering Lecturer at Computers and Systems Engineering
Department, Department,
Faculty of Engineering Mansoura University, Egypt Faculty of Engineering Mansoura University, Egypt

Mohamed S. Elksasy Faiz F. Areed

Assist. Prof at Computers and Systems Engineering Prof at Computers and Systems Engineering
Department, Department,
Faculty of Engineering Mansoura University, Egypt Faculty of Engineering Mansoura University, Egypt

ABSTRACT step, one phase is positive energized, and one is negative

Brushless DC (BLDC) motors are one of the most interesting energized, and the third phase is floating.
motors, not only because of their efficiency, and torque This commutation can be done using Hall sensors or without
characteristics, but also because they have the advantages of using them, which is called sensorless commutation. This
being a direct current (DC) supplied, but eliminating the sensorless commutation depends on detecting the back EMF
disadvantages of using Brushes. BLDC motors have a very in each phase of the stator winding, and so detects the zero
wide range of speed, so speed control is a very important issue crossing point (ZCP) of these back EMF. When a ZCP is
for it. There are a lot of parameters which need to be in focus detected the motor shall be commutated to the next step, so a
while talking about a speed controller performance like smooth continues motion can be achieved.
starting current, starting torque, rise time, etc. There are two
main methods for controlling the speed, PID Controllers, and In [1, 2, 4] the commutation using ZCP detection is discussed.
Fuzzy PI controllers. Both are different in complexity and In [2, 4, 5] the lines voltage differences are used. In [6] the
performance. In this paper, the PI and Fuzzy PI speed position of the rotor is detected using the differences in the
controllers for the BLDC motors will be proposed. A phase's reactance due to their position according to the
simulation study is conducted to evaluate the efficiency of the permanent magnet rotor. In [7] the rotor speed detection using
proposed speed controllers. Further, a comparative study is back EMF and ZCP detection is discussed.
performed to validate the system effectiveness. Speed control is the process of changing the electric power
delivered to the motor to achieve a certain speed which is
Keywords needed. Another purpose of speed control is to keep the motor
BLDC Motor, Speed Control, PI Controller, Fuzzy Controller,
speed constant with outsource disturbances in torque.
Fuzzy PI.
There are two main methods for controlling the speed, the
1. INTRODUCTION Proportional integral derivative controller, and the fuzzy PI
Automation Brushless Direct Current (BLDC) motors are controller. In [3, 8] both conventional PI, and fuzzy PI
used in wide fields and applications, such as Aerospace, controllers are discussed.
Medical equipment, etc. This is because of its high efficiency,
wide speed range and their high torque relative to motor size. In case of motor speed control, using the derivative part
Additionally, not using brushes have the following increase the effect of noise, so the most industrial applications
advantages[4]: didn't use any derivative parts in the controllers, so in most
industrial applications a PI controller is used to come over the
 No sparks during operation which allows using the noise issue and increase the system stability. In this paper the
motor in hazard areas. derivative part gain will be zero, and a PI controller will be
used.
 No noise from commutating sparks.
Although most industrial control systems depend on PI
 Low maintenance cost, and long life as there is no controllers, most of these applications are nonlinear (like
brushes need to be replaced. temperature control), and PI tuning for nonlinear systems is
 Smaller motor size. very difficult [9]. On the other hand, Fuzzy PI controllers can
be used for nonlinear systems, but it need good knowledge of
 No friction at commutator. the system for tuning.
One of the most important advantages of these motors is the In section 2 the conventional PI speed controller for BLDC
speed torque linearity, and the capability of controlling the will be discussed, and fuzzy PI speed controller will be
speed by changing the applied DC voltage. BLDC motors, presented in section 3. A comparative study will be presented
like any other motor, consist of a rotor and a stator. But in for both PI controller, and fuzzy PI controller in section 4.
brushless dc motors the rotor is a permanent magnet and the Finally a comparison between PI speed controller and fuzzy
stator is a winding distributed, in most cases, in three phases. PI speed controller will be shown in section5.
BLDC motors are six steps electrically commutated, in each

29
 International Journal of Computer Applications (0975 – 8887)
Volume 109 – No. 10, January 2015

2. CONVENTIONAL PI CONTROLLER So any controller, PI controller or fuzzy PI controller, target to

The target from any controller is to minimize the error reduce the rise time, settling time, steady error, and overshoot.
between the actual output, which needed to be controlled, and
the desired output, which is called the set point. In the case of
3. FUZZY PI CONTROLLER
speed control this error can be expressed by the following Fuzzy controller is a logistic controller based on fuzzy logic.
equation: Fuzzy controllers depend on rules and conditions between
inputs to get the output.
e(t) = ω𝑆𝑃 (𝑡) − ω𝑃𝑉 (𝑡) (1)
Fuzzy controllers rules are in terms that human can
Where e(t) is the error function of time, ω𝑆𝑃 (𝑡) is the understand like tall, short, medium height, so it is easier for
reference speed or the speed set point as function of time, and human to design if he has a well knowledge about the system
ω𝑃𝑉 (𝑡) is the actual motor speed as function of time. The PI that needed to be controlled.
term stands for Proportional Integral Derivative, so any PI
The inputs of the fuzzy controller are mapped to certain
controller can be divided into 3 parts each part has its Gain,
values called Fuzzy sets. Any fuzzy controller consists of
the first part is the proportional part witch is the error
three parts (see Figure 2).
multiplied by a constant gain which is KP. The second part is
the integral part, which is the integration of error with time  Fuzzification: the process of converting the
multiplied by a constant gain, which is KI. The third part is analogue input to one of the values of the fuzzy sets
the derivative part, which is the derivative of error with time using a membership function.
multiplied by a constant gain which is KD. The PI controller
equation can be expressed as the following.  Rule Base: Are the logistic rules or conditions
between the inputs to get the output.
u t = K 𝑃 𝑒 𝑡 + 𝐾𝐼 𝑒 𝑡 𝑑𝑡 (2)
 Defuzzification: is the process which convert the
Where u(t) is the PI output, KP is the proportional gain, KI is system output from the fuzzy sets values to
the integral gain, KD is the derivative gain, and e(t) is the analogue output value.
error function shown in equation (1). The following function
block, in figure 1, explains the operation of the PI controller. error
Fuzzification Rule Base
Set point Error PI Out
Change of
∑ PI Controller Process error
ωSP(t) e(t) u(t)
Actual Speed
Output
Defuzzification
ωPV(t)Block Diagram
Figure 1 PI Controller
According to [9], there are four main parameters which should Figure 2 Fuzzy Controller Block Diagram
be minimized by the control system:
Most fuzzy controllers use a rectangular membership function
 Rise time (Tr): defined as the time taken to go from and two fuzzy sets or more can be used. The more fuzzy sets
10% to 90% of the targeted set point value. used the more stability and better performance achieved, but
also more complexity the system becomes. In case of 5 fuzzy
 Settling time (Ts): defined as the time required for sets, the fuzzy sets may be called Negative Big NB, Negative
the response curve to reach and stay within a range N, Zero Z, Positive Bid PB, and Positive P.
of certain percentage (usually 5% or 2%) of the
final value. [10] 4. PI CONTROLLER AND ITS
 Steady state error: defined as the difference between SIMULATION RESULTS.
the steady state output and the desired output. As mentioned above most industrial applications do not use
the derivative part in the PID. In this paper a PI controller is
 Overshooting: defined as the maximum peak value proposed. The BLDC motor parameters are shown in
of the response curve measured from the desired appendix I. The used PI coefficients are as follows:
response of the system [11].Overshooting is the KP = 0.0014 and KI = 0.17.
maximum value in the response curve minus the
targeted value divided by the targeted value, and At the beginning the speed is zero, and at 0.012 sec the speed
can be calculated as shown in equation 3. reference increases to 700 RPM, at 0.1 sec a load of 0.1 NM is
Max .Speed Response − Setpoint
added, and finally at 0.2 sec the speed reference is increased
Overshooting = (3) to 900 RPM. Figure 3 shows the used simulation model in
Setpoint
Simulink for the PI speed controller.
There are more parameters which should be taken into
account in case of motor speed control, like start up current,
start up torque, and speed variation percentage.

30
 International Journal of Computer Applications (0975 – 8887)
Volume 109 – No. 10, January 2015

In figure 6, the speed variations is less than 0.8%, and at

0.1sec when the 0.1 NM load is added, the speed changed to
680 RPM, and returned back to its steady value after 0.03 sec.

Figure 6 Speed Variations.

In figure 7, where the BLDC motor phases currents are
shown, the start-up current is about 6 A. At 0.2 sec, where the
set point changed to 900 RPM, the current rise up to2 A, and
then returned to its steady value.

Figure 3 Used Simulink Model

In figure 4, the speed response is shown, the rise time at the
first step is 0.005 sec, and there is no overshoot, the settling
time is about 0.025 sec.

Figure 4 Speed Response using PI

For the second step (from 700 to 900 RPM), the rise time is
0.005 sec, and there is no overshoot, the settling time is about
0.0075 sec. From figure 5, where the torque response is
shown, the start-up torque is about 7.9 N.M. At 0.2 sec, where
the set point changed to 900 RPM, the torque rise up to
2.2N.M, and then returned to its steady value.

Figure 7 Stator Current in Phase a, b, and c

5. FUZZY PI CONTROLLER AND

SIMULATION RESULTS
The proposed fuzzy PI controller use 5 fuzzy sets, NB, N, Z,
P, and PB. And two inputs, the first input is the error, which is
defined by the following equation.
E(N) = ω𝑆𝑃 (𝑁) − ω𝑃𝑉 (𝑁) (4)
Where N = 1, 2, 3, etc. And E(N) is the error at sample N,
ω𝑆𝑃 (𝑁) is the reference speed or the speed set point at sample
Figure 5 Torque Response using PI N, and ω𝑃𝑉 (𝑁) is the actual motor speed at sample N.

31
 International Journal of Computer Applications (0975 – 8887)
Volume 109 – No. 10, January 2015

The second input is the change of error, which is defined by Figure 9 shows the simulation Simulink model used for
the following equations. simulating the proposed Fuzzy PI controller.
CE(N) = E(𝑁) − E(𝑁 − 1) (5)
Where N = 1, 2, 3, etc. And CE(N) is the change of error at
sample N, E(N) is the error at sample N, E(N - 1) is the error
at sample N – 1. The membership function is a triangular
function, as shown in figure 8.

Figure 9 Fuzzy PI controller Simulink model

From figure 10, the speed response is shown, the rise time at
the first step is0.02 sec, and there is no overshoot, the settling
time is about 0.023 sec. In figure 11, the torque response is
shown. The start-up torque is about 2.1 N.M. At 0.2 sec,
where the set point changed to 900 RPM, the torque rise up
to2.3 N.M, and then returned to its steady value.

Figure 8 Fuzzy Membership Function

To optimize the fuzzy PI controller performance, a gain factor
is multiplied by the error, and another one multiplied by the
change of error, and another one is multiplied by the output,
so the membership function range can be changed easier.
These factors are 0.15 for the error, 15 for the change of error,
and 0.003 for the output.
Table 1 shows the rule base of the fuzzy PI controller used,
there are 25 rule used from the 5 fuzzy sets of the two inputs.
Table 1 Fuzzy PI Controller Rule Base
Figure 10 Speed Response using Fuzzy PI controller
E NB N Z P PB
CE

NB NB NB NB N Z

N NB NB N Z P

Z NB N Z P PB

P N Z P PB PB

PB Z P PB PB PB Figure 11 Torque response using Fuzzy PI Controller

32
 International Journal of Computer Applications (0975 – 8887)
Volume 109 – No. 10, January 2015

In figure 12, the speed variations is less than 0.6%, and at In [3], in case of PI controller, the settling time in the first step
0.1sec when the 0.1 NM load is added, the speed changed to (from 0 to 700 RPM) is 0.06455 sec. and in case of Fuzzy PI
690RPM and returned back to its steady value after 0.01 sec. controller, the settling time is 0.05972 sec.
Table 2: Comparative Study

Reference [3]

speed Conventional PI Controller

Tr Ts

0-700 0.04 0.06455

700-900 0.005 0.007

Figure 12 Speed Variations
In figure 13, where the BLDC motor phases currents are speed Fuzzy PI Controller
shown, the start-up current is about 2 A. At 0.2 sec, where the
set point changed to 900 RPM, the current rise up to about 0-700 0.039 0.05972
2A, and then returned to its steady value.
700-900 0.01 0.018

Proposed Controller

Speed Conventional PI Controller

0-700 0.005 0.025

700-900 0.005 0.0075

Speed Fuzzy PI Controller

0-700 0.02 0.023

700-900 0.004 0.005

The other performance parameters are extracted from the

response speed curve of both the PI controller and the Fuzzy
PI controller. From the table above, it is clear that in the first
step (from 0 to 700 RPM), the proposed PI Controller
performance is better, as the proposed one has a very small
rise time, which is 0.005sec. And it is increased in stability
without oscillations.
The proposed PI controller has also a small settling time,
which is 0.025 sec with almost no overshoot. For the second
step (from 700 to 900 RPM), the performance of the proposed
PI controller is almost the same as the compared one, and the
speed is increased in stability too.
The start-up torque in the proposed PI controller is about
7.9N.M in the first step, but in [3] the start-up torque is about
3N.M.

Figure 13 Motor Phases Current The start-up current in the proposed PI controller is about 6A,
while in [3], the start-up current is 4 A. so in [3], a smaller
6. COMPARATIVE STUDY start-up current is achieved successfully, but with low start-up
In [3] a PI speed controller and a fuzzy PI speed controller is torque.
used to control the speed of a BLDC motor with the same
From the fuzzy PI controller, it is clear that in the first step
parameters of the one used in this paper. Table 2 shows a
(from 0 to 700 RPM), the proposed fuzzy PI Controller
comparison between the performance in both papers for both
performance is better, as the proposed one has a very small
the PI controller and Fuzzy PI controller.
rise time, which is 0.02 sec. and small settling time is
0.023sec with no overshoot.

33
 International Journal of Computer Applications (0975 – 8887)
Volume 109 – No. 10, January 2015

For the second step (from 700 to 900 RPM), the performance characteristics. Also by current control, the speed and torque
of the proposed fuzzy PI controller is almost better than the variations can be reduced to minimum, by avoiding any
compared one, as the rise time is about 0.004 sec, and settling sudden changes in the motor current value.
time is 0.005 sec.
APPENDIX I
The start-up torque in the proposed fuzzy PI controller is The Parameters of the BLDC Motor used in this Paper.
about 2.2 N.M in the first step, but in [3] the start-up torque is
about 0.9781 N.M
Rated speed 3000 RPM
The start-up current in the proposed fuzzy PI controller is
about 2A, while in [3], the start-up current is 1 A. Number of Poles 4
7. PI CONTROLLER AND FUZZY PI Resistance/Phase (ohm) 10.91ohm
CONTROLLER PERFORMANCE
COMPARISON Inductance/Phase(H) 30.01e-3H
Table 3 contains a detailed comparison between the proposed
PI controller and the proposed Fuzzy PI controller. Moment of inertia(kg/m2) 2.ge-4 kg/m2
Table 3: Comparative Study between PI Controller and
Fuzzy PI Controller. Voltage constant
136.1357
(V_peak L-L / krpm)
PI Fuzzy PI
Parameter Torque constant(Nm/A) 1.3
Controller Controller

0-700 RPM 0.005 0.02 9. REFERENCES

Rise Time [1] Li Chunfeng, Yu Weibo, Tang Wusheng "Study on
Rotor Position of Sensorless Brushless DC Motors
700-900 RPM 0.005 0.004
Through Back Electromotive Force Detection" presented
at E-Product E-Service and E-Entertainment (ICEEE)
0-700 RPM 0.025 0.023 2010 International Conference on, 7-9 Nov. 2010.
Settling
time [2] Xiong hui, Xue YanBo "The Design of Brushless DC
700-900 RPM 0.0075 0.005
Motor Back-EMF control" presented at Environmental
Science and Information Application Technology
0-700 RPM 0% 0.7% (ESIAT) 2010 International Conference on, 17-18 July
Steady state
error 2010.
700-900 RPM 0% 0.7%
[3] Madhusudan Singh , Archna Garg "Performance
Evaluation of BLDC Motor with Conventional PI and
0-700 RPM 7.9 N.M 2.1 N.M
Start-up Fuzzy Speed Controller " presented at
torque 10.1109/IICPE.2012.6450516 Conference: Power
700-900 RPM 2.2 N.M 2.3 N.M Electronics (IICPE), 2012 IEEE 5th India International
Conference on.
0-700 RPM 6A 2A
Start-up [4] Padmaraja Yedamale "Brushless DC (BLDC)
Current Motor Fundamentals" Microchip Technology Inc,
700-900 RPM 2A 2A AN885 application note, 11/11/2003. Available:
http://ww1.microchip.com/downloads/en/AppNotes/0088
Speed 5a.pdf.
0.8% 0.6%
Variations
[5] Yu-Jhen Fu, Lu-Ting Huang, Ruey-Yue Lin, and Cheng-
Hu Chen "Reliable Starting Method for Sensorless
8. CONCLUSION AND FUTURE WORK Brushless DC Motor Drive" presented at Advanced
A BLDC motor speed controller is presented in this paper, Intelligent Mechatronics (AIM) 2012 IEEE/ASME
using both PI controller, and Fuzzy PI controller. International Conference on, 11-14 July 2012.
A comparative study has been done between the presented PI [6] Yen-Chuan Chang and Ying-Yu Tzou "A New
controller and Fuzzy PI controller, and the used references to Sensorless Starting Method for Brushless DC Motors
evaluate the presented speed controllers' performance. In without Reversing Rotation" presented at Power
general the presented speed controllers have better Electronics Specialists Conference, 17-21 June 2007.
performance. A comparison between the presented PI speed PESC 2007. IEEE.
controller and the fuzzy PI speed controller has been done too,
and it is clear that the performance of the fuzzy PI speed [7] M. Baszynski, and S. Pirog "A novel speed measurement
controller is in general better than the PI speed controller. method for a high-speed BLDC motor based on the
signals from the rotor position sensor" IEEE, 30 January
A future work could be done to add current control function to 2013 10.1109/TII.2013.2243740.
the proposed speed controller, so the current can be kept
within a certain range for a given speed, which will help in [8] Sreekala.P, and Prof.Dr.A.Sivasubramanian "Speed
enhancing the motor startup current, reducing the motor Control Of Brushless Dc Motor With Pi And Fuzzy
current ripples, and enhancing the motor torque Logic Controller Using Resonantpole Inverter" presented

34
 International Journal of Computer Applications (0975 – 8887)
Volume 109 – No. 10, January 2015

at Innovative Smart Grid Technologies - India (ISGT [10] Tay, Teng-Tiow; Iven Mareels; John B. Moore (1997).
India), 2011 IEEE PES, 1-3 Dec. 2011. High performance control. Birkhäuser. p. 93.ISBN
0-8176-4004-5.
[9] R. Arulmozhiyal, and R.Kandiban "An Intelligent Speed
Controller for Brushless DC Motor" presented at [11] Ogata, Katsuhiko (1987). Discrete-time control systems.
Industrial Electronics and Applications (ICIEA), 2012 Prentice-Hall. p. 344. ISBN 0-13-216102-8.
7th IEEE Conference on, 18-20 July 2012.

IJCATM : www.ijcaonline.org 35