Speed Tracking of Field Oriented Control Permanent Magnet Synchronous Motor Using Neural Network

International Journal of Power Electronics and Drive System (IJPEDS)
Vol. 4, No. 3, September 2014, pp. 290~298
ISSN: 2088-8694  290
Journal homepage: http://iaesjournal.com/online/index.php/IJPEDS
Speed Tracking of Field Oriented Control Permanent Magnet
Synchronous Motor Using Neural Network
Wahyu Mulyo Utomo*, Nooradzianie Muhammad Zin*, Zainal Alam Haron*, Sy Yi Sim*, Azuwien
Aida Bohari*, Roslina Mat Ariff**, Dirman Hanafi**
* Department of Power Electrical, Universiti Tun Hussein Onn Malaysia
** Department of Mechatronic and Robotic, Universiti Tun Hussein Onn Malaysia
Article Info ABSTRACT
Article history:
Received Mar 9, 2014
Revised Apr 28, 2014
Accepted May 15, 2014
The field oriented control theory and space vector pulse width modulation
technique make a permanent magnet synchronous motor can achieve the
performance as well as a DC motor. However, due to the nonlinearity of the
permanent magnet synchronous motor drive characteristics, it is difficult to
control by using conventional proportional-integral-derivative controller. By
this reason in this paper an online neural network controller for the
permanent magnet synchronous motor is proposed. The controller is
designed to tracks variations of speed references and also during load
disturbance. The effectiveness of the proposed method is verified by develop
simulation model in MATLAB-simulink program. The simulation results
show that the proposed controller can reduce the overshoot, settling time and
rise time. It can be concluded that the performance of the controller is
improved.
Keyword:
Permanent Magnet
Synchronous Motor
Neural Network
Copyright © 2014 Institute of Advanced Engineering and Science.
All rights reserved.
Corresponding Author:
Wahyu Mulyo Utomo,
Department of Power Electrical,
Universiti Tun Hussein Onn Malaysia
86400 Parit Raja, Batu Pahat, Johor, Malaysia
Email: wahyu@uthm.edu.my
1. INTRODUCTION
Permanent magnet synchronous motors (PMSM) has been use either in low or mid power
applications such as adjustable speed drives, computer peripheral equipment, robotics and electric vehicles
lately. Permanent magnet synchronous motor has the features density of high power, free maintenances and
high efficiency which has been used in widely application in the various electric drives [1]. In 1998, Pillay &
Krishnan presented that PM motor drives and classified them into two categories which are permanent
magnet synchronous motor drives (PMSM) and brushless dc motor (BDCM) drives [2]. The PMSM has a
sinusoidal back emf and to produce the constant torque, it need sinusoidal stator currents while the BDCM
has a trapezoidal back emf and to produce the constant torque, it need rectangular stator currents. The PMSM
has features similar to a wound rotor synchronous machine except the permanent magnet of PMSM will
produce excitation instead of field winding for servo applications and tend to not have any damper windings.
The PM motor family divided into two categories which are internal rotor and external rotor. Both designs
are used in critical applications like wind power generators and elevator winches.
There are two methods to achieve zero steady state error: switching and integration. To eliminate
steady state error, a Proportional- Integral (PI) controller should be employed [3]. By using PI controller
exact dq axis reactance parameters can be obtained. Moreover, to step change of command speed, parameter
variations and load disturbances is very sensitive. Since it is slightly simple to implement, Proportional-
Integral-Differential (PID) controller become most widely used for PMSM. So, a real time self automated
intelligent hardware implementation of PID controller as well as FOC is desired [4].

IJPEDS ISSN: 2088-8694 
Speed Tracking of Field Oriented Control Permanent Magnet Synchronous Motor… (Wahyu Mulyo Utomo)
291
The PM motors can be used in Vector Control (VC) or so called Field Oriented Control applications
[5]. In order to drive a PMSM smoothly, an Electronic Control Unit (ECU) shall be designed such that the
stator current space vector. So, FOC can achieve this goal. In FOC, motor stator currents & voltages are
manipulated in the direct-quadrature (d-q) reference frame of the
rotor.
The artificial neural networks (ANN) are best suited for solving the problems that are nonlinear in
nature. In ANN we can use parallel processing methods to solve some real-world problems where it is
difficult to define a conventional algorithms. The ability of ANN to learn large classes of nonlinear functions
is well known [7]-[8]. It can be trained to emulate the unknown nonlinear plant dynamics by presenting a
suitable set of input/output patterns generated by the plant [6]. Once system dynamics have been identified
using an ANN, many conventional control techniques can be applied to achieve the desired objective.
In this paper, a model of ANN closed-loop PMSM control system that is controlled by SVPWM are
develops and presents an analysis of ANN speed controller for speed performance in FOC PMSM drive. The
effectiveness of the proposed method is verified by develop simulation model in MATLAB-simulink.
2. RESEARCH METHOD
2.1. Permanent Magnet Synchronous Motor Dynamic Modeling
PMSM is essentially a three phase AC motor with sinusoidal back EMF driven by a DC source,
which is converted to three-phase alternating currents supplying to the three stator windings of PMSM. The
mathematic model of PMSM idq synchronous rotating reference frame can be obtained from synchronous
machine model. Due to the constant field produced by permanent magnets, the field variation is zero. It is
also assumed that saturation and losses of core are negligible, the induced emf is sinusoidal and there is no
damper winding on rotor. Using these assumptions, the voltage equations can write as follow:
d s d d d q e q
d d
v R i L i L i
dt dt
   (1)
q s q q q d e d e PM
d d
v R i L i L i
dt dt
      (2)
The produced torque of the machine can be presented as follow:
3
[ ( ) ]
2
PM q d q d qeT P i L L i i   (3)
While, the maximum speed can be identified from the relationship:
L f m me
d
T T K J
dt
    (4)
The update frequency of the control loops must be high enough and the SVPWM should be properly
configured to ensure sinusoidal currents applied to the stator windings.
2.2. Field Oriented Controller (FOC) Description
Field oriented control (FOC) also known as decoupling or vector control, came into the field of ac
drives research in the late 1960s and was developed prominently in the 1980s to meet the challenges of
oscillating flux and torque response in inverter fed induction and synchronous motor drive. The inexplicable
dynamic behavior of large current transients and the resulting failure of inverters was a curse and barrier to
the entry of inverter fed ac drives into the market.
FOC is a process for handling motor control resulting in energy-efficient operation and fast dynamic
response at all speeds. It commutates motor by calculating the voltage and current vector based on motor
current feedback. It maintains high efficiency in a wide operating range and allows precise dynamic control
of speed and torque. The FOC control the stator currents represented by a space vector. It transforms the
three-phase stator currents (a, b, c) into the two-phase system variants (α, β) . A two time invariant coordinate
system (d-q) is obtained from the system variants. For this system, d (direct) part is making the motor flux
while q part (quadrature) is generate the toque. In FOC, motor stator currents and voltages are manipulated in

 ISSN: 2088-8694
IJPEDS Vol. 4, No. 3, September 2014 : 290 – 298
292
directquadrature (d-q) reference frame of the rotor, a way must be mathematically transformed using the Park
and Clarke transformations before they can be used to output SVPWM. Which means that the stator current
feedback transformed from the three-phase static reference frame of the stator windings to the two axis
rotating d-q references frame of the rotor.
2.3. Permanent Magnet Synchronous Motor Drive System
Figure 1 is the diagram of velocity/current control loop using FOC technology based on proposed
ANN speed controller and PID speed controller.
Figure 1. PMSM drive system with PID controller
This paper proposed an ANN control method of FOC based on SVPWM to reduce the overshoot,
steady state error and rise time. The ANN speed control is added to the speed controller to produce the speed
reference. The block diagram of the proposed ANN speed controller of FOC for PMSM drive system is
shown in Figure 2.
Figure 2. PMSM drive system with ANN controller
2.3. Proposed ANN Speed Controller Structure
To design the neural network control some information about the plant is required. Basically, the
numbers of input and output neuron at each layer are equal to the number of input and output signals of the
system respectively. Further the number of hidden layers and the total neurons is depended on the complexity
of the system and the required training accuracy [9]. To implement search efficiency optimal control of
PMSM drive, a multilayer perceptron neural network control is developed. Based on the type of the task to be
performed, the structure of the proposed ANN speed controller is shown in Figure 3[10].

293
Figure 3. PMSM drive system with ANN controller
The controller consists of input layer, hidden layer and output layer. Based on number of the neuron
in the layers, the ANN is defined as a 1-3-1 network structure. The first neuron of the output layer is used as
a torque reference signal (a2
1=mf). The connections weight parameter between jth
and ith
neuron at mth
layer is
given by wm
ij, while bias parameter of this layer at ith
neuron is given by bm
i. Transfer function of the network
at ith
neuron in mth
layer is defined by:
1
1
1
m
S
m m m m
i ij j i
j
n w a b



  (5)
The output function of neuron at mth
layer is given by:
( )m m m
i ia f n (6)
where: f is activation function of the neuron.
In this design the activation function of the output layer is unity and for the hidden layer is a tangent
hyperbolic function given by:
2
2
( ) 1
1
m
i
m m
i n
f n
e
 

(7)
Updating of the connection weight and bias parameters are given by:
( )
( 1) ( )m m
ij ij m
ij
F k
w k w k
w


  

(8)
( )
( 1) ( )m m
i i m
i
F k
b k b k
b


  

(9)
After the neural network architecture is modelled, the next stage defines the learning model to
update network parameters. By this learning capability, it makes the ANN suitable to be implemented for the
system with motor parameters which are difficult to define and vary against with environment. The training
process minimizes the error output of the network through an optimization method. Generally, in learning
mode of the neural network controller a sufficient training data input-output mapping data of a plant is
required. Since the motor parameters of the PMSM drive vary with temperature and magnetic saturation, the
online learning Back propagation algorithm is developed. Based on first order optimization scheme, updating
of the network parameters are determined. The performance index sum of square error is given by:
21
( ) ( )
2
i
i
F k e k  (10)
( ) ( ) ( )i i ie k t k a k  (11)
where: ti is target signal
ai output signal on last layer.

 ISSN: 2088-8694
294
The gradient descent of the performance index against to the connection weight is given by:
m
nF F i
m m m
w n wij i ij
 

  
(12)
The sensitivity parameter of the network is defined as:
m
i m
i
F
s
n



(13)
m
m i
i m m
i i
aF
s
a n


 
(14)
Gradient the transfer function again to the connection weight parameter is given by:
1
m
mi
im
ij
n
a
w



(15)
From substitution Equation (13) and (15) into (8) the updating connection parameter is given by:
1 1
( 1) ( ) ( ) ( )m m i m m
ij i i iw k w k s k a k  
   (16)
With the same technique the updating bias parameter is given by:
1
( 1) ( ) ( )m m i m
i i ib k b k s k 
   (17)
3. RESULTS AND ANALYSIS
The simulation result of the PMSM drive system is presented to verify the feasibility of the
proposed ANN speed controller under various operating speed conditions. In this section the dynamic model
of a three-phase PMSM, space vector PWM and neural network control model have been developed. The
simulation is developed using Borland C++, and then embedded as S-function in Simulink-Matlab. The
parameters for the motor are given in Table 1.
The simulation is observed during start up response which is by using constant references speed 100
rad/s with no load condition as shown in Figure 4. With the same reference speed, the simulations of both
PID and ANN speed controllers are run simultaneously.
Table 1. Parameters of PMSM
Variables Parameter Value
P Poles 4
J Moment of Inertia 0.0008 [kgm²]
Ld,Lq
Rs
λpm
d-q axis Inductance
Armature Resistance
Flux
0.0085 [H]
2.875 [Ω]
0.175 [Wb]

295
Figure 4. Start up response for ANN and PID speed controller
From the results, it show that by using ANN speed controller produced a better start-up performance
compare to the PID speed controller where the overshoot is totally removed and the settling time more faster
than PID speed controller in achieving desired output speed . Second testing is observed during stepping-up
and stepping-down response application. For stepping- up response, the speed reference is varying from
60rad/sec to 120rad/sec is shown in Figure 5.
Figure 5. Stepping-up response for ANN and PID speed controller
For stepping-down response, the speed reference is varying from 120rad/sec to 40rad/sec is shown in Figure
6.
Figure 6. Stepping-up response for ANN and PID speed controller

 ISSN: 2088-8694
296
The last testing, the simulation is verify by using triangular speed reference for both PID and ANN speed
controllers as shown in Figure 7.
Figure 7. Output speed when using triangular speed reference
Both system PID and ANN speed controller is also tested on the effect of nominal load disturbance.
Figure 9 shows the load disturbance applied and Figure 8 shows the speed tracking response for both
systems. The system is tested with value of 5 Nm load at 1.0s with 80rad/sec speed reference.
Figure 8. Load disturbance for PID and ANN speed controller

297
Figure 9. Output speed when load disturbance is applied
Refer to the Figure 8, at the initial both systems are running with no load. From Figure 9, it is
obvious that the ANN controller drop the speed from 80rad/sec to 79rad/sec with great settling time. While,
PID controller drop the speed from 80rad/sec to 71.3rad/sec with largest settling time. So, it show that that the
proposed ANN speed controller produce significant improvement control performance compare to the PID
controller such as reduce overshoot, settling time and rise time speed response.
4. CONCLUSION
This paper has presented the modeling and simulation of the field oriented control for PMSM drive
using online neural network controller. In order to investigate effectiveness of the proposed controller, some
reference speed model was tested. The simulation study is realized in MATLAB program. The results of the
proposed and conventional PID controller are plotted in the same speed graph with intention to make details
comparison based on visual observation on steady state error, rise time and overshoot speed responses. It can
be conclude that the speed performance of the PMSM can be improved by applying the online ANN
controller scheme.
ACKNOWLEDGEMENTS
All the authors would like to express a sincere acknowledgment to Universiti Tun Hussein Onn
Malaysia for the valuable support during completion this research and manuscript.
REFERENCES
[1] K Song, W Liu, G Luo. Fuzzy Logic Based Online Electromagnetic Loss Minimization of PMSM Drive. IEEE
Transactions on Vehicle Power and Propulsion. 2008: 1.
[2] P Pillay, R Krishnan. Modeling, Simulation, and Analysis of Permanent-Magnet Motor Drives, Part 1: The
Permanent-Magnet Synchronous Motor Drive. IEEE Transactions on Industry Applications. 1989; 25: 265-273.
[3] LK Wong, FHF Bung, PKS Tam. Combination of Sliding Mode Controller and PI Controller using Fuzzy Logic
Controller. IEEE International Conference on Fuzzy System. 1998; 1: 296-301.
[4] M Marufuzzaman, MBI Reaz, MA Mohd Ali. FPGA Implementation of an Intelligent Current dq PI Controller for
FOC PMSM Drive. IEEE Transactions on Computer Applications and Industrial Electronics (ICCAIE). 2010: 602.
[5] ES Sergaki, SP Georgilakis, AG Kladas, GS Stavrakakis. Fuzzy Logic Based Online Electromagnetic Loss
Minimization of PMSM Drives. IEEE Transactions on Electrical Machines. 2008; 1.
[6] J Weidong, W Qunjing, C Quan, S Xiaofeng. SVPWM Strategy for Three-Level Inverter based on SVPWM
Strategy for Two-Level Inverter. Transactions of China Electrotechnical Society. 2009; 24: 108-114.
[7] Y Yi, D Mahinda, MA Rahman. Implementation of an Artificial Neural Network based Real Time Adaptive
Controller for an Interior PMSM. IEEE Transaction On Industry Application. 2003; 39: 96-103.
[8] KS Narenda, K Parthasarathy. Identification and Control of Dynamical Systems using Neural Networks,"IEEE
Transaction Neural Network. 1990: 4-27.
[9] I Choy, SH Kwon, JY Choi, JW Kim, KB Kim. On-Line Efficiency Optimization Control of a Slip Angular
Frequency Controlled Induction Motor Drive Using Neural Networks. IECON Proceedings 13 annual Conference.
1996; 1216-1221.
[10] AHM Yatim, WM Utomo. Online Optimal Control of Variable Speed Compressor Motor Drive System using
Neural Control Model. IEEE Transaction On Power and Energy. 2004; 83- 87.

 ISSN: 2088-8694
298
BIOGRAPHIES OF AUTHORS
Wahyu Mulyo Utomo was born in Pati, Indonesia, in 1969. He received the B. S Degree in
electrical engineering from the Universitas Brawijaya Malang, in 1993, the M.S. degree in
electrical engineering from the Institute Sepuluh Nopember Surabaya, in 2000, and the Dr. Eng.
Degree from Universiti Teknologi Malaysia in 2007. He is currently a Senior Lecturer in Electrical
Power Engineering Department, Faculty of Electrical and Electronic Engineering,
Universiti Tun Hussein Onn Malaysia. His current research interests include the area of power
electronics and motor drive control.
Nooradzianie Muhammad Zin was born in Kedah, Malaysia, in 1987. She received the B.S. degree
in electrical engineering from Universiti Tun Hussein Onn Malaysia, in 2011. She is currently a
Master student in the Electrical Power Engineering Department , Faculty of Electrical and
ElectronicEngineering, Universiti Tun Hussein Onn Malaysia. His current research interests
include the area of power electronics and motor drives control.
Zainal Alam Haron was born in Malacca, Malaysia, in 1956. He received the B.S. degree in
electrical engineering from the University of Glasgow, U.K., in 1981, the M. S Degree in
Ionisation Physics from the University of Wales, U.K., in 1985, and the Dr. Eng. Degree from
University of Newcastle Upon Tyne, U.K., in 1992. He is currently an Associate Professor in the
Electrical Power Engineering Department, Faculty of Electrical and Electronic Engineering,
electronics and electric motor drives.
Sim Sy Yi was born in Johor, Malaysia, in 1988. She received the B.S. degree in electrical
engineering from Universiti Tun Hussein Onn Malaysia, in 2011. She is currently a PhD student in
the Electrical Power Engineering Department , Faculty of Electrical and ElectronicEngineering,
electronics and motor drives control.
Azuwien Aida Bohari was born in Johor, Malaysia, in 1987. She received the B.S. degree in
electrical engineering from Universiti Tun Hussein Onn Malaysia, in 2011. She is currently a
Master student in the Electrical Power Engineering Department , Faculty of Electrical and
ElectronicEngineering, Universiti Tun Hussein Onn Malaysia. His current research interests
include the area of power electronics and motor drives control.
Roslina Mat Ariff was born in Perak, Malaysia, in 1987. She received the B.S. degree in electrical
engineering from Universiti Tun Hussein Onn Malaysia, in 2011. She is currently a PhD student in
the Electrical Power Engineering Department , Faculty of Electrical and ElectronicEngineering,
electronics and motor drives control.
Dirman Hanafi The author was born in Agam Regency, West Sumatera, Indonesia, in 1967. He
received the B. S Degree in electrical engineering from Universitas Bung Hatta, Padang, Indonesia,
in 1994, the M.S. degree in electronic engineering from the Institute Technology Bandung, in 1997,
and the Dr. Eng. Degree from Universiti Teknologi Malaysia in 2006. He is currently a Senior
Lecturer in Mechatronics and Robotics Engineering Department, Faculty of Electrical and
Electronic Engineering, Universiti Tun Hussein Onn Malaysia. His current research interests
include the area of Intelligent System identification and Intelligent Control System.

Speed Tracking of Field Oriented Control Permanent Magnet Synchronous Motor Using Neural Network

More Related Content

What's hot (20)

Similar to Speed Tracking of Field Oriented Control Permanent Magnet Synchronous Motor Using Neural Network (20)

More from IJPEDS-IAES (20)

Recently uploaded (20)

Speed Tracking of Field Oriented Control Permanent Magnet Synchronous Motor Using Neural Network