2018 International Conference on Computer, Control, Informatics and its Applications

Angular Position Control of Brushless DC Motor

Using Cuckoo Search algorithm
Optimized PD Controller
Rizqi Andry Ardiansyah, Edwar Yazid
Research Centre for Electrical Power and Mechatronics
Indonesian Institute of Sciences
Bandung, West Java, Indonesia
[email protected], [email protected]

Abstract—Controlling the angular position of brushless DC by the concept of PID controller in Section 3. Concept of
(BLDC) motor is an essential task to improve the transient and Cuckoo search combined with PD controller is put on
the steady state performances under loaded condition. Many Section 4. Section 5 describes the experimental setup and
problems in BLDC controlling, one of them is the overshoot so Section 6 presents the experimental results and discussions.
that the angular position control of BLDC motor using cuckoo
Finally, some conclusions are written in Section 7.
search algorithm optimized proportional derivative (PD)
controller to form what the so-called the cuckoo-PD (Cuckoo-
PD) controller is proposed in this paper. Control system is II. MATHEMATICAL MODEL OF A BLDC MOTOR
realized in a microcontroller namely a 16MHz Atmega 2560 BLDC motor studied in this paper is three-phase type
with an absolute encoder as a position sensor. The effectiveness and depicted in Fig. 1. It is produced by Oriental Motor
of the Cuckoo-PD controller is tested under constant and
Corp. with the type of KC AXHM5100 Vexta. Electronic
varying step functions with the Ziegler-Nichols-PD (ZN-PD)
controller as a benchmark. Experimental results show that the circuit of such a motor can be drawn in Fig. 2.
Cuckoo-PD controller has slower speed than the ZN-PD
controller, but the latter has overshoot and small oscillations
during its steady state condition as a consequent of having a
fast rise time.

Keywords—angular position, BLDC motor, Cuckoo search,

PD controller

I. INTRODUCTION
Tuning the gains of proportional-derivative (PD)
controller for controlling the brushless DC (BLDC) motor
using nature-inspired metaheuristic algorithms is still an
open research in the control engineering field. Prior research
can be tracked from several published works. Premkumar
and Manikandan [1] proposed bat algorithm for tuning the
gains PID controller. It is applied to control the varying Fig.1. BLDC motor [12]
speed and load conditions of BLDC motor. Their results
show that the proposed controller is superior to the classical
PI, fuzzy tuned PID controller and fuzzy variable structure
controller. Ibrahim et al. [2] conducted a comparative study Q1 Q2
Q3

between particle swarm optimization (PSO) and bacterial

BLDC Motor
foraging (BF) algorithms for tuning the gains of PID Phas e 1
controller in order to control the speed of a BLDC motor via Phase 2
Phas e 3
L1
simulation. Their results shows that the PSO algorithm has L2 L3
Q5 Q6
better control performance. Other soft computing algorithms Q4

such as fuzzy logic and neural network for tuning the gains
of PD controller can be found in references [3]-[11],
respectively. Many research in the field of BLDC
controlling with many method has been done, but the research
Fig. 2 Electronic circuit of Fig. 1
still needs to overcome the problems. This research is proposed to
improve the control capacity of PD controller in the field of BLDC
motor controlling. By taking the three-phase current is at balanced condition as
The rest of this paper is organized as follows: Section 2 in (1), then general voltage equation of Fig. 2 can be
derives the mathematical model of a BLDC motor, followed expressed in a state-space as in (2),

978-1-5386-5741-6/18/$31.00 c 2018 IEEE 152

Authorized licensed use limited to: Turkish Energy Nuclear and Mineral Research Agency. Downloaded on April 10,2025 at 09:38:10 UTC from IEEE Xplore. Restrictions apply.
 i x + i y + i z = 0. (1) By using technical specifications shown in Table 1 [12], and
respective chosen parameters, control simulation of BLDC
motor can be performed and compared to the experimental
ux  R 0 0 ix   L M M ix  ex 
           (2) work as described in the next section.
uy  = 0 R 0 iy  + M L M iy  + ey  .
u  0 0 R i  M L L  i  e 
 z    z    z  z III. CUCKOO SEARCH ALGORITHM OPTIMIZED PD CONTROLLER
Equation (1) contains parameters such as: phase coils Optimization process of the gains of PD controller,
voltage and phase current from coils stator voltage which K p and K d through Cuckoo search algorithm is depicted
are denoted by u x , u y , u z and i x , i y , i z , respectively. in Fig. 3. Let say, θ and θ d denote the actual and reference
Other parameters are back EMF, phase resistant, and phase angular position of BLDC motor in discrete time index n ,
inductance which are denoted by e x , e y , e z , R, and L, respectively, then the angular position error e θ and the
respectively. The mutual inductance of the three phase error derivation Δ e θ can be calculated by,
winding is M (H). By utilizing the mechanical angular
speed, and the parameters above, electromagnetic torque eθ ( n ) = θ d ( n ) − θ ( n ) . (6a)
can be calculated as,
ex ix + ey iy + ez iz Δe θ ( n ) = θ d ( n ) − θ ( n − 1 ) . (6b)
Te = . (3)
ωm Error and its derivative are then multiplied with the
Mechanical torque can be then derived by employing the
( )
proportional gain K p , integral gain ( K i ) and derivative
moment inertia, J and friction coefficient, B and expressed gain (K d ) to form the control action (u n ) which is sent to
in (4), driver of BLDC motor. Formulation of control action can be
dω m completely expressed in (7),
Tm = J + B ω m . (4)
dt
u (n ) = K p e (n) + K d Δ e T ( n ) . (7)
By combining (3) and (4), the rate of angular speed can be
written as, Equation (7) suggests that under any types of system
input for θ d , the gains in (7) significantly affect the control
d ω m Te −Tm B
= − ω m . (5) response, θ of closed-loop system in Fig. 3. Suboptimal
dt J J
gains lead to the control response becomes unstable, high
Table I. KC AXHM5100 BLDC motor specification overshoot and large steady-state error. Both gains must be
tuned to match the system input (set-point) and the system
Parameter Value output by giving the corrective action in terms of control
action so that high position accuracy can be achieved.
Rated Output Power - HP(W) 1/8(100)
Hence, Cuckoo algorithm is proposed to tune both gains.
Voltage - VDC 24 +/- 10% For the sake of clarity, optimizer for the parameters of
Rated Input Current - A 6 PDC using Cuckoo search algorithm is called CS-PD. The
Rated Torque (N·m) 0.40 major phases of this algorithm are depicted in Fig. 4. Both
Permissible Load Inertia J (10-4 kg·m2) 5.6
{ }
gains are denoted as X = K p , K d , initiated randomly
Speed max (RPM) 3000 and are called as cuckoo to form a number of nests. Each
cuckoo evaluates the nests using following cost function,
N
SSE =  (θ d ( n ) − θ ( n ) ) . (8)
n =1

θ d (n) + u (n) u T (n )
+ θ (n)
eθ (n)
BLDC
Motor
−
+
Δeθ (n) PDC

Cuckoo Search

Fig. 3. Diagram block of proposed PD controller for BLDC

153
Authorized licensed use limited to: Turkish Energy Nuclear and Mineral Research Agency. Downloaded on April 10,2025 at 09:38:10 UTC from IEEE Xplore. Restrictions apply.
 Cost Function, Eq. (8)
Cuckoo

e KP
Updating Mechanism
Eq. (9)
Δe KD
Gains
Cuckoo

Fig. 4. Gains optimization process in CS-PD controller

This objective function is sufficient for controlling the The internal 30 pulses per revolution rotary encoder
angular position of BLDC motor. The nests are then updated sensor is chosen to measure the motor speed. Before
using following updating mechanism [13], conducting the experiment, the connection of BLDC motor
to its driver unit must be ensured. Five wire signal pins
(t +1) (t )
= X i + α ⊕ δ ∗ L evy (λ ) + (1 + δ )
' (Start/Stop, Run/Brake, CW/CCW, EXT Inp, Speed Output)
Xi
from BLDC motor driver unit is then connected to five pins
(9) of microcontroller. External interrupt and timer feature in
 R  P (t )- X (t )  + R  P (t )- X (t )   .
 1 2 i  Atmega 2560 are used to measure the motor speed. In order
  i i   g 
to record the motor into a database of a computer, a serial
Equation (9) suggests that L evy (λ ) is the key parameter in
'
communication is used to send the speed data from
Cuckoo search algorithm. This function is a random walk microcontroller to computer. Sampling rate for the
' experiment is 50 ms.
L evy distribution which controls the distance between the
(t )
best nest of individual historical P i and the best nest of
(t )
global P g with its current nest. Equation (9) contains
other four parameters: α, δ , R1 and R 2 in which α is a
constant parameter related to step size and equal to 1, δ is a
weight index in the interval [ 0, 1] while R1 and R 2 are two
random numbers uniformly also selected from the
interval [ 0, 1] . Updating the nests is repeated until the stop
criterion (maximum iteration number or minimum error
value) is satisfied. Equation (9) also suggests that Cuckoo (a) (b)
search algorithm is not a complicated optimization
technique as it involves only one updating mechanism.

IV. EXPERIMENTAL SETUP

Experimental work is conducted in mechatronic division
laboratory, Research Centre for Electrical Power and
Mechatronics, Indonesian Institute of Science. Setup for
real-time control implementation for the proposed Cuckoo-
PD controller is shown in Fig. 5. Two main electronic
circuits are used in this experiment, namely a 16MHz
Atmega 2560 microcontroller unit and a BLDC motor driver (c) (d)
unit. Three-phases AXH Series BLDC motor with Fig. 5. Experimental setup (a) BLDC motor (b) BLDC
24Volt/100 watt from Oriental motor is taken as a plant. driver circuit (c) computer (d) microcontroller
Main controller is a 16MHz Atmega 2560 microcontroller
unit which is used to drive the BLDC motor as well as to
V. RESULTS AND DISCUSSIONS
record the motor angular position. The maximum voltage is
24 Volt and is limited up to 5 Vref with Pulse Width In this section, performance of proposed CS-PD
Modulation Signal. controller is investigated under two cases, constant and
varying step functions. Control performances in time
domain are assessed in terms of rise time, settling time,
overshoot and peak time and compared. Parameters used in

154
Authorized licensed use limited to: Turkish Energy Nuclear and Mineral Research Agency. Downloaded on April 10,2025 at 09:38:10 UTC from IEEE Xplore. Restrictions apply.
 the CS-PD controller are given by 50 nests with a weight
4
index of 0.9. CS-PD controller is executed in 20 iterations
3.5
with the interval for the searching space is 0 ≤ x ≤ 50
for { K P , K D } . Details are tabulated in Table 2. The first 3

case is BLDC motor is commanded to track an angular 2.5

ZN-PD

Volt
position θ d → 30 0
by giving a constant step function. The 2 CS-PD
1.5
control response for time duration of 33 s is shown in Fig.
6(a). It is seen that ZN-PD and CS-PD controllers are 1
capable of tracking the reference angular position and both 0.5
controllers have successfully reached it with respect to time. 0
However, each controller has different performances 0 5 10 15 20 25 30
during the tracking especially for the CS-PD controller as
Time
confirmed in Table 3. ZN-PD controller has faster rise time (b)
than the CS-PD controller in which the ratio is around one
and half. This is due to the value of proportional gain Fig. 6. Constant step function (a) control response (b) control action
produced by ZN-PD controller is higher than that of CS-PD
controller as confirmed by Table 4. High control action Table III. Performances of controllers under constant step function
shown b Fig. 6(b) reconfirms it. However, having fast rise
time leads to experience overshoot of 3.33%, and small Performance ZN-PD CS-PD
oscillations during its steady state condition. Fast settling
time is another superiority of proposing CS-PD controller. Rise time (s) 0.18 0.27
However, peak time for both controllers is almost similar in Settling time (s) 32.67 11.6
this case. Details of performances and the gains of each Overshoot (%) 3.33 0
controller are tabulated in Table 3 and Table 4, respectively. Peak time (s) 11.6 11.12
In optimizing the gains, PD controller optimized by
Cuckoo search algorithm produces a cost function as shown Table IV. Gains of controllers for constant step function
in Fig. 7. It displays the cost function with respect to the
number of iterations. As observed, the cost drops Gain
significantly at the-3rd iteration, and starts to converge after Controller
the-4th iteration to a certain value. Fig. 7 suggests that the P D
optimization process is performed quickly.
ZN-PD 1.67 0.63
CS-PD 0.9 0.03
Table II. Parameters of Cuckoo search algorithm
Cuckoo Search
Number of 200
20 Weight index 0.9
generations
Number of nests 50 '
Distribution L evy
Cost function

Number of cuckoo 2 Constant parameter 1 150

100

50
1 5 10 15 20
Number of iterations

Fig. 7. Cost function for constant step function

The second case is BLDC motor is commanded to track

a varying angular position θ d →300 ,400 ,500 ,200 ,400 by giving
a varying step function. Response and control actions for
time duration of 81 s are shown in Fig. 8. Again, the BLDC
35
Time motor is able to track the reference angular position, and
30 (a) both controllers have successfully tracked it with respect to
time. Finding results in constant step function are found for
25 Reference this case. In the time duration of 0-30 s, ZN-PD controller
20
ZN-PD has faster rise time than GA-PD controller, where it is 0.08 s
Ɵ (deg)

CS-PD
15

10

5
155
0
Authorized licensed use limited to: Turkish Energy Nuclear and Mineral Research Agency. Downloaded on April 10,2025 at 09:38:10 UTC from IEEE Xplore. Restrictions apply.
0 5 10 15 20 25 30
 against 0.175 s. However, the overshoot is 22.22 % against 5 400
%. This reconfirms the advantage of using CS-PD controller.
350

300

Cost function
250

200

150

100

50
2 4 6 8 10 12 14 16 18 20

Number of iterations
Fig. 9. Cost function for varying step function
50
Time (s)
Reference
ZN-PD (a) Under varying step function, PD controller optimized by
40 CS-PD Cuckoo search algorithm produces a cost function as shown
in Fig. 9. Similar with constant step function, the cost drops
Ɵ (deg)

30 significantly at the-3rd iteration, and starts to converge after

the-12th iteration to a certain value.
20
VI. CONCLUSION
10
In this paper, Cuckoo search optimized PD controller
0 (CS-PD) controller is proposed for controlling the angular
0 10 20 30 40 50 60 70 80 position of BLDC motor in a real-time implementation.
Experimental results show that the proposed controller can
improve the performance of BLDC motor in tracking the
6
given angular positions in terms of constant and varying
step functions. Important conclusions and suggestions from
ZN-PD Time (s) this paper are derived below.
5
CS-PD (b)
- The CS-PD controller makes the tuning process of the
4
Fig. 8. Varying step function (a) control response (b) control action gains of PD controller more practical.
Volt

3
- Under constant step function, ZN-PD controller has
Table V. Controller’s performances under varying step function faster rise time than CS-PD controller with overshoot of
2 3.33%, and small oscillations. The latter has faster
Performance ZN-PD CS-PD settling time than the earlier.
1 - Under varying step function, ZN-PD controller
Rise time (s) 0.08 0.17 experiences overshoot of 22.22 %, where CS-PD
0 controller experiences overshoot of 5 %.
0 Settling
10 time 20
(s) 30 2.69
40 50 60 0.69
70 80
Overshoot (%) 22.22 5.00
Peak time (s) 0.30 1.70 ACKNOWLEDGMENT
Research facility for this work from Research Centre for
Electrical Power and Mechatronics, Indonesian Institute of
Sciences is gratefully appreciated.
Table VI. Gains of controllers for varying step function
REFERENCES
Gain [1] K. Premkumar, and B.V. Manikandan, “Speed control of BLDC
Controller
motor using bat algorithm optimized adaptive Neuro-Fuzzy Inference
P D System,” Applied Soft Computing, vol. 138, pp. 260-270, 2014.
[2] H.E.A. Ibrahim, F.N., Hassan, and A.O. Shomer, “Optimal PID
ZN-PD 1.58 0.4 control of a brushless DC motor using PSO and BF techniques,” Ain
CS-PD 0.85 0.02 Shams Engineering Journal, vol. 5, pp. 391-398, 2014.
[3] Akash Varshney, Deeksha Gupta, Bharti Dwivedi, “Speed response
of brushless DC motor using fuzzy PID controller under varying load
condition”, Journal of Electrical Systems and Information
Technology, vol. 4 ,pp. 310–321, 2017.
[4] Chang Hyun Kim, Ji Hoon Yang, Dong Gyun Lim, Byung Suhl Suh,
“An Enhanced PID Controller for Speed Control of Brushless DC

156
Authorized licensed use limited to: Turkish Energy Nuclear and Mineral Research Agency. Downloaded on April 10,2025 at 09:38:10 UTC from IEEE Xplore. Restrictions apply.
 Motors Based on Convex Set Optimization”, IFAC Proceedings ,vol. Araz Darba,Frederik De Belie,Pieter D'haese,Jan A. Melkebeek, “Improved
43, issue 22,, pp. 75-80, 2010. Dynamic Behavior in BLDC Drives Using Model Predictive Speed and
R.Kandiban, R.Arulmozhiyal, “Speed Control of BLDC Motor Using Current Control”, IEEE Transactions on Industrial Electronics, vol. 63, pp.
Adaptive Fuzzy PID Controller”, Procedia Engineering, vol. 38, pp. 306- 728 – 740, 2016.
313, 2012. Arman Jaya,Era Purwanto, Melinda Badriatul Fauziah,Farid Dwi
E.Gowthaman,VVinodhini,MirYasserHussain,S.K.Dhinakaran,TSabarinath Murdianto,Gigih Prabowo,Muhammad Rizani Rusli, “Design of PID-fuzzy
an, “Speed Control of Permanent Magnet Brushless DC Motor Using for speed control of brushless DC motor in dynamic electric vehicle to
Hybrid Fuzzy Proportional plus Integral plus Derivative Controller”, improve steady-state performance”, IES-ETA 2017, pp. 179-184, 2017.
Energy Procedia, vol. 117, pp. 1101-1108, 2017. Premkumar, K. and Manikandan, B.V., “Fuzzy PID Supervised Online
Silviu Dumitriu, “Intelligent Dsp-Based Motion Control Of A Brushless ANFIS Based Speed Controller for Brushless DC Motor”,
D.E. Motor”, IFAC Proceedings, vol. , pp. 319-324,2000. Neurocomputing, vol. 157, pp. 76-90, 2015.
Jino Joya, S. Ushakumarib, “Performance Comparison of a Sensorless KC AXHM5100 BLDCM user manual, Oriental Motor, 2014.
PMBLDC Motor Drive System with Conventional and Fuzzy Logic Zhigang Lian, Lihua Lu, Yangquan Chen, A New Cuckoo Search,
Controllers”, Procedia Technology, vol. 25, pp. 643 – 651, 2016. International Conference on Intelligence Science ICIS 2017: Intelligence
Science, pp. 75.

157
Authorized licensed use limited to: Turkish Energy Nuclear and Mineral Research Agency. Downloaded on April 10,2025 at 09:38:10 UTC from IEEE Xplore. Restrictions apply.