Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

2nd IEEE International conference on power Electronics, Intelligent Control and Energy systems (ICPEICES-2018)

Design of FPI-PD Controller for Brushless DC Motor


Roshan Bharti, Rishika Trivedi, Prabin Kumar Padhy
Department of Electronics and Communication Engineering,
PDPM Indian Institute of Information Technology, Desgn and Manufacturing, Jabalpur, India
1
bhartiroshan99@gmail.com
2rishika.t@iiitdmj.ac.in
3prabin16@iiitdmj.ac.in

Abstract— In this paper, a new fuzzy proportional-integral- advance controllers. There are various kind of methods
proportional-derivative (FPI-PD) controller is designed, in which available for PID controller tuning [5],[6] like Ziegler Nichols,
fuzzy PI controller is in the forward path and conventional PD Cohen-Coon, Error-Trial, Chien-Hrones-Reswick, Good gain
controller is in inner feedback loop. The gain parameters of the method, Pole placement method, Kappa-tau method, and
proposed controller are calculated by gradient descent
Optimization method etc., but among them, Ziegler-Nicholas
optimization method. The performance evaluation of this
controller is performed for speed control of brushless DC motor in [7] is most popular and widely used. The problem with
MATLAB/Simulink. The obtained responses are compared with conventional PID controller is, it gives a better result only when
conventional PID controller and the existing fuzzy PID controller. the parameters are well matched with the system dynamics.
It reveals that proposed FPI-PD controller is more efficient than However, obtaining the system dynamics of a nonlinear plant is
both the above mentioned controllers. a challenging task. Therefore, intelligent controllers like
Artificial Neural Network (ANN) and the fuzzy logic based
Keywords— PID, FLC, FPI-PD, Brushless DC motor, controllers are widely used for controlling nonlinear plants. The
Gradient descent optimization method drawback of using ANN is that, it requires input-output data for
training purpose [8]. Besides, fuzzy logic does not require exact
I. INTRODUCTION information about plant model like zeros and poles of the
Electrical motors are used in almost all electromechanical transfer function.
movement around us. It converts electrical energy into
mechanical energy. There are two types of motors: AC and DC In 1965, Lotfi A Zadeh proposed the fuzzy set theory
type. The conventional DC motor is used in various applications [9]. In 1974, E.H Mamdani introduced a first fuzzy logic
like electric shaver, trimmer, hair dryer, cranes, starting motor controller for laboratory build steam engine [10]. Fuzzy logic
in the car and two-wheeler, photocopy or Xerox machine etc., controller essentially contains four parts: - fuzzification, fuzzy
due to its simple design and low cost [1]. In conventional DC inference system, fuzzy rule base and defuzzification. The
motor, brushes are used for the commutation, which is made up advantage of the fuzzy logic controller is that it can be easily
of carbon. The brushes tear out after a span of time which leads incorporated with a conventional controller like PI, PD and PID
to sparking and eventually damages the motor. The frequent and this combination performance better than a conventional
maintenance of conventional DC motor is required from time to controller. The major issue with Fuzzy logic based controller is
time to avoid wear and tear of the brushes. This is the main the tuning of the gain parameters. Since, the fuzzy logic
reason due to which conventional DC motor is not used for the controller is nonlinear, so, tuning of the gain parameters is
applications which requires durability and hence, the use of crucial as compared to PID controller. If number of gain
brushless DC motor is increasing day by day. Brushless DC parameters are less then linear approximation [11] can be used
(BLDC) motor is extensively used in numerous applications for gain calculation. It is difficult to obtain large number of gain
like heating and ventilation (HVAC), servo robotic positioning parameters using linear approximation. Consequently,
actuators, traction, fans, quadcopter, transportation, washing optimization techniques used to calculate the gain parameters of
machine, mixer, disc drives, printers, blowers, and drones. The the fuzzy incorporated controller. Various literatures have
advantages of BLDC motor are low maintenance due to reported different optimization techniques for the tuning of the
unavailability of brushes, high reliability, high power to weight fuzzy-based controller such as, Genetic algorithm [12]. Particle
ratio, high efficiency, high speed and electronic control [2]. swarm optimization [13], Firefly algorithm [14], Bat algorithm
BLDC motor is a type of permanent magnet synchronous motor [2], Cuckoo search Algorithm [15], Hybrid bacterial-foraging
which is driven by DC voltage via three phase inverter but swarm optimization [16] etc.
commutation of current is achieved by electronic switches. The
instant of commutation is depended on the position of the rotor In this paper, a new FPI-PD controller is proposed for
which is determined by a speed sensor or sensor-less techniques the speed control of BLDC motor where a fuzzy PI controller is
[3]. The back EMF in BLDC motor is of two types either in forward path and a conventional PD controller is in inner
sinusoidal or trapezoidal [4]. feedback loop. The gain parameters of the proposed controller
Due to a simple structure and low cost, PID controller are calculated by gradient descent optimization method [17].
is widely used in industries even after availability of many The mathematical model of BLDC motor is derived for

978-1-5386-6625-8/18/$31.00 ©2018 IEEE


864
evaluating the performance of the proposed controller. τ m = 17.1ms = 0.0171 s
MATLAB/Simulink environment is used for the simulation. R = 1.20 Ω ,
The obtained results are then compared with conventional PID
J = 92.5 gcm 2 = 9.25 × 10 −6 Kgm 2 ,
controller and the existing fuzzy PID controller. The results
show that proposed controller performs better than both of the mNM
kt = 25.5 = 25.5 × 10 −3 Nm / A
above-mentioned controllers in terms of rise time, percentage A
overshoot and settling time. 9.25 × 10−6 × 3 × 1.20 (8)
ke = = 0.0763 v − secs / rad .
II. MATHEMATICAL MODELLING OF BRUSHLESS DC MOTOR 0.0171× 25.5 × 10−3
The transfer function of BLDC motor is obtained by: The transfer function of Maxon motor can be obtained
1 by putting the value of ke , τ m and τ e in equation (1):
ke (1) 1 (9)
G ( s) = ,
τ m .τ e .s 2 + τ m .s + 1 G (s) = 0.0763 ,
0.0171× 155.56 × 10−6.s 2 + 0.0171.s + 1
where, τ m is mechanical time constant and τ e is electrical time
or,
constant (10)
13.11
R.J J. R G (s) = .
τm =  = , (2)
−6
2.66 × 10 .s 2 + 0.0171.s + 1
ke .kt ke .kt
and,
L L (3) III. CONVENTIONAL PID CONTROLLER
τe =  = .
R R PID controller considered is the parallel combination of three
Since, there is a three-phase symmetrical arrangement control modes, namely, proportional, integral and derivative as
in brushless DC motor, so mechanical and electrical time shown in Fig. 1. It has combined feature of both PD and PI
constant will become: controller. It improves both transient as well as steady state
J .3R (4) performance of the plant.
τm = ,
ke .kt
and,
L (5)
τe = ,
3R
where,
ke is back EMF constant in v-secs/rad and kt is torque constant
in N-m/A.

The specifications used for modelling of this BLDC


motor is obtained from datasheet of EC 45 Ø45 mm, brushless, Fig. 1:PID Controller
30-watt Maxon motor [18].
The output equation of PID controller is:
TABLE I d (11)
CHARACTERISTICS OF BLDC MOTOR U PID ( t ) = K p e ( t ) + K i  e ( t ) dt + K d e (t )
Maxon motor data Value
dt
or,
Terminal resistance (per phase) 1.20 Ω
 1  (12)
Terminal inductance (per phase) 0.560 mH U PID ( s ) = K p e ( s ) 1 + + sTd  ,
Rotor inertia  Ti. .s 
92.5 gcm 2
where, e, Td , Ti . , K p , K d and Ki represent feedback error,
Torque constant 25.5 mNm/A
derivative time, integral time, proportional gain, derivative gain
Mechanical time constant 17.1 ms
and integral gain, respectively.
From equation (5):
0.560 × 10−3 (6) IV. FUZZY PID CONTROLLER
τe = = 155.56 × 10−6 . The architecture shown in Fig. 2 is one of the most popular
3 × 1.20
From equation (4), ke can be obtained: architecture of fuzzy PID controller (FPID). There are two
inputs: error (e) and change in error (ce). There is one output u
J .3R (7) of FPID controller, which is used for self-tuning the parameters
ke = ,
τ m .kt of PID controller. The inputs and the output are divided into
where, seven linguistic variables using five triangular and two gaussian

865
membership functions. The membership functions for both shown in Table II. The linguistic variables are PL, PM, PS, Z,
inputs and output is shown in Fig. 3 and Fig. 4, respectively. NS, NM and NL, namely, positive large, positive medium,
positive small, zero, negative small, negative medium and
negative large, respectively.

TABLE II
RULE BASE FOR FPID CONTROLLER [19]
ce /e NL NM NS Z PS PM PL
NL PL PL PL PL PL PS Z
NM PL PM PM PM PS Z NS
NS PL PM PS PS Z NS NM
Z PM PM PS Z NS NM NS
PS PM PS Z NS NM NM NL
Fig. 2: Fuzzy PID Controller [19] PM PS Z NS NM NM NL NL
PL Z NS NM NL NL NL NL

V. PROPOSED FPI-PD CONTROLLER

A. Controller Architecture
The proposed structure of FPI-PD controller is shown in Fig. 5.
In this structure, fuzzy PI controller is connected in the forward
path and conventional PD controller is connected in the inner
feedback loop. In this proposed structure, the fuzzy logic
controller has two inputs: normalized error (E) and normalized
change in error (CE). It has two outputs u1 and u2 which are
Fig. 3: Membership function for inputs e and ce
used to adjust the proportional and integral gain of PI controller,
respectively. The inputs are divided into seven linguistics
variable using seven triangular membership functions and the
outputs are divided in to five linguistic variables using five
triangular membership functions. The rule base for FPI
controller is shown in Table III, where S, M, L, VL, PL, PM,
PS, Z, NS, NM, and NL are known as small, medium, large,
very large, positive large, positive medium, positive small, zero,
negative small, negative medium, and negative large
respectively. The membership functions for inputs and outputs
are shown in Fig. 6 and Fig. 7, respectively. In this controller,
Fig. 4: Membership function for output u. Mamdani model is used for fuzzy inference system and Centre
of gravity (COG) method is used for defuzzification
In this FPID controller, Mamdani model is used for There are six scaling factors ( k1 − k6 ) in this proposed
fuzzy inference system and Centre of gravity (COG) method is controller, where ( k1 − k2 ) , ( k3 − k4 ) and ( k5 − k6 ) are the
used for defuzzification. The rule base for FPID controller is
scaling factors of the inputs, PI controller and PD controller,

Fig. 5. Proposed FPI-PD controller structure

866
respectively. The calculation of scaling factors ( k1 − k6 ) is a The iteration process has repeated according to the
following equation:
difficult task because it is committed with both: fuzzy PI and
X K +1 = X K + λK* S K = X K − λK* ∇f K (13)
PD controller. Therefore, gradient descent optimization
technique is used to calculate the scaling factors of this where, is search direction, K is a number of iteration, λK* is
proposed controller. optimal step size and ∇f K is gradient vector of function.
Si = −∇f K = −∇f ( X K ) (14)

Fig. 5: Membership function for inputs E and CE.

Fig. 6: Membership function for outputs u1 and u2.

TABLE III
RULE BASE FOR FPI CONTROLLER [16]
CE/E NL NM NS Z PS PM PL
NL VL VL VL VL VL VL VL
NM L VL VL VL VL VL L
NS M L VL VL VL M M
Z S M L VL L L S
PS M L VL VL VL M M
PM L VL VL VL VL VL L
PL VL VL VL VL VL VL VL

B. Gradient descent optimization method:


In 1847 it was Cauchy who first minimized a function by taking
negative of the gradient vector as a search direction. The
Fig. 7: Flowchart of gradient descent optimization method
gradient descent method which is also known as steepest
descent method[20] is first order iterative optimization VI. SIMULATION RESULTS
algorithm which is used to determine the local minimum of the
function. In gradient descent optimization method we start from The obtained transfer function of BLDC motor is used for
an initial point ( ) and iteratively move towards the direction simulation. The simulation is run for 0.05 seconds by
of negative of gradient vector until the minimum point is considering the fixed step size of 0.00001. The gain parameters
obtained. The flow chart of gradient descent optimization of the proposed controller are calculated by gradient descent
method is shown in Fig. 8. Similarly, to find local maximum of optimization method via Simulink design optimization toolbox
the function one needs to consider steps equivalent to positive in MATLAB/Simulink. There are two set-point operating
of the gradient of that function. This method is named as condition is considered. In the first case, there is a fixed speed
gradient ascent method. in which a load disturbance of 0.3 is given at 0.03

867
seconds. In the second case, a variable speed is considered in TABLE V
PERFORMANCE INDICES OF CONTROLLERS FOR BLDC MOTOR AT FIXED SPEED
which speed goes down from 1 to 0.5 at 0.025 seconds. The
WITH LOAD DISTURBANCE
responses of brushless DC motor for both the set-point Controller Rise time % Overshoot Settling time
operating condition is shown in Fig. 9 and Fig.10, respectively.
TABLE IV (in ms) (in ms)
GAIN PARAMETERS OF CONTROLLERS FOR BLDC MOTOR
11.327 PID 2.340 22.619 16.446
k P
PID Controller 1381.34 FPID 0.296814 13.068 0.7907
kI
(Ziegler-Nicholas Method) 0.0232
kD FPI-PD 0.059822 6.471 0.148016

k1 43.7774
FPI-PD Controller 1.8678 TABLE VI
k2 PERFORMANCE INDICES OF CONTROLLERS FOR BLDC MOTOR AT VARIABLE
(Gradient descent optimization 338.1774
method) k3 SPEED
Controller Rise time % Overshoot Settling time
k4 0.2554
(in ms) (in ms)
k5 0.0025

k6 0.0062 PID 3.744 45.352 22.50

FPID 0.168017 25.556 1.1269

FPI-PD 0.042049 11.224 0.156923

VII. CONCLUSION
The performance of proposed FPI-PD Controller is evaluated
by controlling the speed of BLDC motor in
MATLAB/Simulink. The mathematical model of BLDC motor
is derived by its parameters. Its speed is controlled by
conventional PID controller, FPID controller and proposed FPI-
PD controller by considering the different set-point operating
condition such as the fixed speed with load disturbance and
variable speed condition. The performance of all the above
mentioned controllers is summarized in Table V and Table VI
for both the set-point operating condition in terms of transient
performance. The results reveal that the proposed FPI-PD
controller outperforms in terms of rise time, percentage
overshoot and settling time, etc.
REFERENCES
Fig. 8: Response of BLDC motor at fixed speed with load disturbance [1] U. K. Bansal and R. Narvey, “Speed Control of DC Motor Using
Fuzzy PID Controller,” vol. 3, no. 9, pp. 1209–1220, 2013.
[2] K. Premkumar and B. V. Manikandan, “Bat algorithm optimized
fuzzy PD based speed controller for brushless direct current motor,”
Eng. Sci. Technol. an Int. J., vol. 19, no. 2, pp. 818–840, 2016.
[3] M. A. Shamseldin and A. A. El-samahy, “Speed control of
BRUSHLESS DC motor by using PID control and self-tuning fuzzy
PID controller,” 15th Int. Work. Res. Educ. Mechatronics, pp. 1–9,
2014.
[4] K. Tabarraee, J. Iyer, S. Chiniforoosh, and J. Jatskevich, “Comparison
of brushless DC motors with trapezoidal and sinusoidal back-EMF,”
in 2011 24th Canadian Conference on Electrical and Computer
Engineering(CCECE), 2011, pp. 000803–000806.
[5] S. B. Prusty, K. K. Mahapatra, U. C. Pati, and S. Padhee,
“Comparative performance analysis of various tuning methods in the
design of PID controller,” Michael Faraday IET Int. Summit 2015, p.
8 (6 .)-8 (6 .), 2015.
[6] P. Cominos and N. Munro, “PID controllers: recent tuning methods
and design to specification,” IEE Proc. - Control Theory Appl., vol.
149, no. 1, pp. 46–53, Jan. 2002.
[7] B. J. G. Ziegler and N. B. N. Rochester, “Optimum Settings for
Automatic Controllers,” 1942.
[8] M. A. H. Azman, J. M. Aris, Z. Hussain, A. A. A. Samat, and A. M.
Nazelan, “A Comparative Study of Fuzzy Logic Controller and
Fig. 9: Response of BLDC motor at variable speed

868
Artificial Neural Network in Speed Control of Separately Excited DC [15] P. Kumar, S. Nema, and P. K. Padhy, “Design of Fuzzy Logic based
Motor,” no. November, pp. 24–26, 2017. PD Controller using cuckoo optimization for inverted pendulum,” in
[9] L. a. Zadeh, “Fuzzy sets,” Inf. Control, vol. 8, no. 3, pp. 338–353, 2014 IEEE International Conference on Advanced Communications,
1965. Control and Computing Technologies, 2014, pp. 141–146.
[10] E. H. Mamdani, “Application of fuzzy algorithms for control of [16] A. Fereidouni, M. A. S. Masoum, and M. Moghbel, “A new adaptive
simple dynamic plant,” Proc. Inst. Electr. Eng., vol. 121, no. 12, p. configuration of PID type fuzzy logic controller,” ISA Trans., vol. 56,
1585, 1974. pp. 222–240, 2015.
[11] R. Manikandan, A. Arulprakash, and R. Arulmozhiyal, “Design of [17] C. C. Soon, R. Ghazali, H. I. Jaafar, and S. Y. S. Hussien, “PID
equivalent fuzzy PID controller from the conventional PID controller tuning optimization using gradient descent technique for an
controller,” 2015 Int. Conf. Control Instrum. Commun. Comput. electro-hydraulic servo system,” J. Teknol., vol. 77, no. 21, pp. 33–
Technol. ICCICCT 2015, pp. 356–362, 2016. 39, 2015.
[12] Zhihong Xiu and Guang Ren, “Optimization design of TS-PID fuzzy [18] O. J. Oguntoyinbo, “PID Control of Brushless DC Motor and Robot
controllers based on genetic algorithms,” in Fifth World Congress on Trajectory Planning Simulation with MATLAB/Simulink,” Thesis
Intelligent Control and Automation (IEEE Cat. No.04EX788), vol. 3, Bachelor, p. 90, 2009.
pp. 2476–2480. [19] A. Varshney, D. Gupta, and B. Dwivedi, “Speed response of
[13] Z. Huang and Y. Wang, “Design of a PSO Based Fuzzy Logic brushless DC motor using fuzzy PID controller under varying load
Controller for Vessel Mooring Shifting System,” in 2008 Fifth condition,” J. Electr. Syst. Inf. Technol., vol. 4, no. 2, pp. 310–321,
International Conference on Fuzzy Systems and Knowledge 2017.
Discovery, 2008, pp. 306–310. [20] S. S. Rao, Engineering Optimization: Theory and Practice. 2009.
[14] R. Trivedi, P. K. Padhy, S. K. Jain, and R. Base, “Design of Fuzzy
PID Controller Using Modified Firefly Algorithm,” Power, Commun.
Inf. Technol. Conf. (PCITC), 2015 IEEE, 2015.

869

You might also like