Brushless DC Motor Drive during Speed Regulation with Artificial Neural Network Controller

Sakshi Solanki Int. Journal of Engineering Research and Application www.ijera.com
ISSN : 2248-9622, Vol. 6, Issue 6, ( Part -3) June 2016, pp.01-05
www.ijera.com 1 | P a g e
Brushless DC Motor Drive during Speed Regulation with
Artificial Neural Network Controller
Sakshi Solanki*
(Department Of Power System Engineering, Galgotias University, Greater Noida)
ABSTRACT
Brushless DC motor, at this moment is extensively used being many industrial functions due to the different
features like high efficiency and dynamic response and high speed range. This paper is proposing a technology
named as Artificial Neural Network controller to control the speed of the brushless DC motor. Here the paper
contributes an analysis of performance Artificial Neural Network controller. Because it is difficult to handle by
the use of conventional PID controller as BLDC drive is a nonlinear. Through PID controller, the speed
regulation of BLDC is not efficient and reliable as PID controller cannot operate the large data, results it gives
different variation in BLDC motor control. The ANN easily trains the data of large amount by NN toolbox. As
ANN controller has the strength to indulge characteristics of control and it is accessible to operate the huge
amount of data as like human can store in a mind. The empirical results prove that an ANN controller can better
control the act than the PID controller. The modelling, control and the simulation of the BLDCM get done by
applying MATLAB/SIMULINK software kit.
Keywords: BLDC motor or BLDCM, ANN Controller, Proportional Integral Derivative (PID) Controller, NN
Toolbox.
I. INTRODUCTION
BLDC motor is consistently determined as
a Permanent Magnet Synchronous motor along a
trapezoidal back EMF waveform shape. The word
“Brushless DC motor” is commonly identifies the
sequence of machines of AC, position sensor of the
rotor, and solid-state inverter that outcomes in a
drive system having characteristics of linear torque.
BLDC motor is only one type of motor which is
attaining demands. BLDC motors in others
situations changes conventional dc motors. Against
the term, BLDC motors are genuinely a type of
permanent magnet synchronous motors. They are
directed by DC voltage although the current
commutation is executed by the solid state
switches. The commutation prompts are tenacious
by the rotor position and the position of the rotor is
revealed either by position sensors or by Sensor
less techniques.
BLDC motors includes many advantages
by conventional DC motors like long operating life,
dynamic response is high, efficiency is high, speed
vs. torque characteristics is improved, operation of
noiseless, speed range is high, torque-weight ratio
is also high. The rise in the energy cost urges
higher appeals of variable speed PM motor drives.
Further, new rapid generation of motor drives into
the automobile industry, established on the hybrid
drives, develops a serious need for great efficient
PM motor drives, and this was the origin of
importance in the BLDC motors. The BLDC is an
electronically controlled commutation system,
rather including a mechanical commutation, which
is characteristic of brushed motors. Also, the
electromagnets do not move the rotation of the
permanent magnets and the armature leftover static.
The problem earned by this that how to shift the
current to a moving armature. According to do this,
the brush system/commutator assembly is restored
by an intelligent electronic controller, which
executes the same power distribution like a brushed
DC motor. Because of the BLDC moves without
brushes its lifetime used can be raised and
operation of maintenance also evade.
II. THE STUDY OF BLDCM DRIVE
The search of BLDC motor is placed on the
acceptance for simplification and windings are
identical and the self and Mutual inductances are
stable. The energetic equations of BLDC motor
using assuming point scan be derived as-
Va = RIa + (L-M)
dia
dt
+ea
Vb =RIb + (L-M)
dib
dt
+eb
Vc =RIc + (L-M)
dic
dt
+ec
Locates, Va, Vb, Vc = Stator phase voltages
Ia, Ib, Ic = Stator phase currents
ea , eb, ec = Back emf of phases
L = Self inductance
M= Mutual inductance
R= Phase resistor
A controller of speed, the braking
chopper, and the intelligent controller models are
limited to the Electric Drives library. It is desirable
RESEARCH ARTICLE OPEN ACCESS

to practise a simplified version of the drive
involving an average-value model of the inverter
for rapid simulation. A Brushless DC motor
(BLDC) is a DC motor twisted inside out; therefore
the field is on the rotor and the armature is on the
stator. The BLDC motor is actually a permanent
magnet AC motor in which a torque-current
tendency mimes the DC motor. Rather of
commutating the armature current by using the
brushes, commutation of electronic is used. This
rejects the problems joined with the brush and the
commutator adjustment for pattern, flickering and
wearing out of the commutator-brush adjustment,
thereby, making a BLDC motor more rugged as
compared to a DC motor. BLDC motors have so
many similarities to the AC induction motors and
the brushed DC motors in terms of the construction
and working principles respectively. Same as other
motors, BLDC motors also consists of a rotor and a
stator.
Figure-1 Speed regulation of BLDC motor
III. SPEED CONTROLLER
 PID Controller
The PID controller is adapted in the
different fields in the engineering payable to its
simplicity, powerful, reliability and tuning
parameters easily. The honoured method to detect
PID parameters is the Ziegler-Nichols rule although
sometimes are not that good in the results.
Unusually, it can be easily accomplished by using
the genetic optimization technique which depends
on three various price functions to detect the better
PID control parameters. The major obstructions
siding PID control technique is rapid change in the
firm point and the parameter deviations, it builds
the PID control which provides a flat response.
This complication can be mitigated by concluding
an advanced control technique such as Artificial
Neural Network Method (ANN).
Figure2- PID controller block diagram
 Artificial Neural Network (ANN) Controller
An ANN-depend controller for the BLDC
motor drive which desires minimal offline training
still absolutely and veraciously pursue command
speed with the insensitivity to load and parameter
deviations. ANNs are the analytical systems
comprising of various weighted interconnected
operation elements that is neurons. The processing
element is an equation, which is usually named as
transfer function. This processing element hold
signals from other neurons and attaches and
disciple them and yields a numerical result. In the
regular, processing elements practically correspond
to the actual neurons, they are attached via a
network and this format establishes neural network.
The network of the ANNs comprises three major
aspect neurons, the connection giving the input and
the output direction, and relation weights indicating
the capability of these relations. Often, the
architecture of an ANN is prepared and weight
values needed to upgrade the certainty of the
outputs are tenacious using one of the different
numerical algorithms. The ANNs resolve a
relationship between the input variables and
approximated variables by certaining the weights
using past examples. In different words, ANNs are
"trained". Formerly these relationships are checked,
an ANN can be regulated with a new data and
evaluations can be generated.
The attainment of a structure is detected
by the calculated signal and error criterion. The
error edge is produced by the observation of the
output of the structure and the calculated output. A
back-propagation algorithm is used to conform the
weights in alike a way to decrease the error edge.
The structure is trained by imitating these
processing bountiful times. The target of training is
too able an optimum solution dependent on the
attainment measurements. ANNs includes a huge
range of functions in actual life problems. They are
presently used fortunately in abounding industries.

Figure3 - ANN Structure
III. SIMULATION RESULTS
PID controller controls the speed of
BLDC motor but not as efficient as Artificial
Neural Network can do. So just to make the motor
more efficient and reliable, we are using the PID
data only for future work. These are too trained for
best results. We have replaced PID controller by
ANN as PID controller have the parameters of the
practical system which changes during operating
conditions. Hence, PID controller failed to give
desired performance under the nonlinearity, load
disturbances, speed regulation of motors. Through
this disadvantage of PID controller, the ANN for its
demand for non linear controllers is purposed here.
Then the inputs (IN1, IN2) and output (OUT) set
for PID controller will be observed in the
workspace of the MATLAB. Then put emfg
algorithm in your current folder so that system can
read all the data. The data provided from the PID is
till 3614.
The emfg is the back emf propogation algotrithm
which is expressed here,
Function y=emfg(u)
if (u>-180)&&(u<=-120)
y=[-1;0;1];
elseif(u>-120)&&(u<=-60)
y=[0;-1,1];
elseif(u>-60)&&(u<=0)
y=[1;-1;0];
elseif(u>0)&&(u<=60)
y=[1;0;-1];
elseif(u>60)&&(u<=120)
y=[0;1;-1];
elseif(u>120)&&(u<=180)
y=[-1;1;0]
end
After retaining this, we secure a set of variables
IN1, IN2 and OUT data. Then in a program
window we type,
“inpt=[IN1.Data IN2.Data]
target=[out.Data]
Then new variables will be created with the name
“inpt” and “target”. And then type in command
window „save data”. By which “data.mat‟ is
created in the current folder.
Finally at the end run, ANN MATLAB code nn1.m
for this data.mat must be in current folder. The
code for nn1.m is as given here,
clc
clear all
close all
load data
net=feedforwardnet(50)
net=int (net);
[net,tr]=train(net,inpt‟,target‟)
Figure4- NN training toolbox
Figure5- Performance Plot

Figure6- Training State Plot
Figure7- Error Histogram Plot
Figure8- Regression Plot
Here, Performance plot shows a plot of training
record error values against the number of training
epochs, Training state plot plots the training state
from a training record returned by train, Error
histogram plots a histogram of error values,
Regression plot displays the network outputs, with
respect to targets for training, validation and test
sets. After getting these plots by training the data,
we will write in a command window „gensim
(net)‟. Then after pressing this command we will
get the ANN simulink block which is blue in color.
Figure9- Simulink block of ANN
Now placing this blue coloured block to
the controller of speed controller by replacing PID
by ANN. As in practical MATLAB, the variation in
the speed of PID and ANN is seen. The speed
variation is given for 1700rpm or 1600rpm.
Therefore it is found that ANN works more
efficiently than PID for the large number of data. In
PID controller if any change occurs, then big
variations occur in the result of speed. And once
ANN gets trained, it gives efficient results.
Figure10- Simulation Block of ANN controller for
speed regulation
IV. CONCLUSION
An Artificial Neural Network controller
(ANN) has been employed for the speed control of
BLDC motor drive and analysis of the results of the
performance of ANN controller is presented. The
modelling and the simulation of the complete drive
system are described in this paper. An
effectiveness of a model is established by the

performance prediction over the wide range of
operating conditions. The performance of the
BLDC motor drive with implication to PID
controller, ANN controller and experimentally
verified with conventional PID controller using
MATLAB/SIMULINK software.
REFERENCES
[1]. B.K. Lee and M. Ehsani, “A simplified
functional model for 3-phase voltage source
inverter using function concept,” IEEE
Trans. On Industrial Electronics, vol.48,
April 2001.
[2]. Yong Liu, Z. Q. Zhu, and David Howe,
“Commutation –Torque –Ripple
minimization in direct –torque-controlled
PM brushless DC drives,” IEEE Trans. On
industry applications, vol. 43, no.4,
July/August 2007.
[3]. Neethu U. and Jisha V.R. , “ Speed control
of brushless dc motor: A comparative study”
2012 IEEE International conference on
power electronics, drives and energy
systems , December 16-19, 2012, Bengaluru,
India.
[4]. Jianwen Shao, Dennis Nolan, Maxime
Teissier and David Swanson, “A novel
microcontroller –based sensorless brushless
dc (BLDC) motor drive for automotive fuel
pumps”, IEEE Trans. On industry
applications, vol.39, no.6,
November/December 2003.
[5]. J.X. Shen, and S.Iwasaki, “Sensorless
control of ultrahigh- speed PM brushless
motor using PLL and third harmonic back
emf”, IEEE Trans. on industrial electronics ,
vol.53, no.2, April 2006.
[6]. P.D.Evans and D.Brown, “Simulation of
brushless DC drives”, Proc. Of the IEEE,
vol.137, no.5, pp. 299-308, September 1990.
[7]. T.J.E. Miller, “Brushless permanent magnet
and reluctance motor drive”, Oxford 1989.

Brushless DC Motor Drive during Speed Regulation with Artificial Neural Network Controller

More Related Content

What's hot (20)

Viewers also liked (19)

Similar to Brushless DC Motor Drive during Speed Regulation with Artificial Neural Network Controller (20)

Recently uploaded (20)

Brushless DC Motor Drive during Speed Regulation with Artificial Neural Network Controller