IRJET-Sound-Quality Predict for Medium Cooling Fan Noise based on BP Neural Network Model

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 01 | Jan-2018 www.irjet.net p-ISSN: 2395-0072
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 980
Sound-Quality predict for medium Cooling Fan Noise Based on Bp
Neural Network Model
Mrs. G. Sangeetha Lakshmi1, Ms. K. Indumathi2, Ms. K. Gowthami 3
1Assistant Professor, Dept of Computer Science and Applications, D.K.M College for Women (Autonomous),
Vellore, Tamilnadu, India.
2,3 Research Scholar, Dept of Computer Science and Applications, D.K.M College for Women (Autonomous),
Vellore, Tamilnadu, India.
-----------------------------------------------------------------------------***----------------------------------------------------------------------------
ABSTRACT: The self-learning capability of this prophecy
method is so strong that each association weight could
generate itself via network learning with a more objective
and accurate predicted result. Base on the complexity and
non-linear description of vehicle cooling fan sound quality
evaluation, the problem about the submission of BP neural
network in sound quality prediction is discussed.
Combining the basic principles and models with examples
the complete implementation process is presented. Then
BP neural network assessment method is compared with
the existing prediction methods, and the conclusion shows
that in interior vehicle subjective and objective noise
evaluation, the prediction effect of the former method is
much better. To a large extent the method could be used to
improve the decision-making accuracy, which is of great
importance to the evaluation, analysis and control of fan
noise nowadays.
Keywords: fan; sound quality; BP neural network;
predict model
I. INTRODUCTION:
An Artificial Neural Network (ANN) is an information
processing paradigm that is inspired by the biological
nervous systems, such as the human brain’s information
processing mechanism. The key element of this
paradigm is the novel structure of the information
processing system. It is composed of a large number of
highly interconnected processing elements (neurons)
working in unison to solve specific problems. An NN is
configured for
A specific application, such as pattern recognition or data
classification, through a learning process. Learning in
biological systems involves adjustments to the synaptic
connections that exist between the neurons. This is true
of NNs as well.
 Neural networks are a new method of programming
computers.
 They are exceptionally good at performing pattern
recognition and other tasks that are very difficult to
program using conventional techniques.
 Programs that employ neural nets are also capable
of learning on their own and adapting to changing
conditions.
Many products and applications employ or include fans,
for example Information Technology (IT) devices and
products, household appliances, air-conditioning
systems and automotive applications.
Low-noise design is a key purchase requirement in all of
these fields, where often the main noise source is a fan.
Forgiven constraints, such as fan dimensions and cooling
performance, it is not possible by technical means to
achieve an arbitrary low-noise design.
The effective characterization of fan noise is challenge in
acoustic and sound quality measurement. In general, the
product sound quality does not depend only on the
emitted noise level or frequency-weighted level like A-
weighted level or loudness. Tonal components, howling
sounds and modulated signals are often the cause of
customer complaints.
Loudness has been introduced as a more hearing related
parameter than the A-weighted level in the last decades.
Several methods exist in international standards for
measuring the loudness of stationary signals (ISO 532 B
and DIN 45631). DIN 45631 will be extended with
respect to loudness of time-varying signals (DIN
45631/A1, draft version published in January 2008).
This is an important step, because in practice we almost
always have to deal with time-varying signals.
Besides time-varying loudness, other psychoacoustic
parameters like sharpness and roughness can be used
for sound quality evaluation. Sharpness considers the
amount of high frequency components of a noise (DIN
45692, draft version published in April 2007), and
roughness evaluates modulation characteristics. In
addition, a metric combining modulation spectral
analysis with loudness calculation has been introduced.
For a long time, researchers have been trying to explore
the establishment of accurate objective quality
parameters to describe people's subjective sense of

sound quality evaluation model. Because of the
randomness of human subjective evaluation and the
existence of some non-linearity between objective
parameters, the existing linear regression model of
sound quality cannot map the complex nonlinear
relationship between subjective and objective
evaluation. To the desired level. Artificial neural network
is a kind of forecasting method with good application
effect, and it is widely used in automobile industry.
Neural network is a adaptive nonlinear dynamic system
composed of a large number of simple neurons and has
many advantages compared to other methods.
This paper starts with the subjective evaluation
experiment and the psychoacoustic parameters metrics.
Then a BP neural network is applied to predict the
radiator fan sound quality.
II. SUBJECTIVE EVALUATION EXPERIMENT:
In the human brain, a typical neuron collects signals
from others through a host of fine structures called
dendrites. The neuron sends out spikes of electrical
activity through a long, thin stand known as an axon,
which splits into thousands of branches. In the training
mode, the neuron can be trained to fire (or not), for
particular input patterns.
In the using mode, when a taught input pattern is
detected at the input, its associated output becomes the
current output. If the input pattern does not belong in
the taught list of input patterns, the firing rule is used to
determine whether to fire or not.
The firing rule is an important concept in neural
networks and accounts for their high flexibility. A firing
rule determines how one calculates whether a neuron
should fire for any input pattern. It relates to all the
input patterns, not only the ones on which the node was
trained on previously.
At the end of each branch, a structure called a synapse
converts the activity from the axon into electrical effects
that inhibit or excite activity in the connected neurons
Pair wise comparison method is a more commonly used
subjective evaluation method, which makes it easy for
the evaluation subject to obtain the relative comparison
result between the evaluation samples. However, when
the number of samples is evaluated, the amount of
evaluation effort required by the paired comparison
method in the general sense will become very large. In
this paper, an improved pair wise comparison method,
which is a group comparison method, which is suitable
for more evaluation samples, is put forward and applied
to the actual subjective evaluation experiment. Group
comparison method can greatly reduce the workload of
subjective evaluation, and make the evaluation
experiment more practical and get more stable
evaluation results.
2.1 Sample collection and processing:
Artificial Neural Networks (ANN) are currently a 'hot'
research area in medicine and it is believed that they will
receive extensive application to biomedical systems in
the next few years.
At the moment, the research is mostly on modeling parts
of the human body and recognizing diseases from
various scans (e.g. cardiograms, CAT scans, ultrasonic
scans, etc.).
Neural networks are ideal in recognizing diseases using
scans since there is no need to provide a specific
algorithm on how to identify the disease.
Neural networks learn by example so the details of how
to recognize the disease are not needed.
What is needed is a set of examples that are
representative of all the variations of the disease.
The quantity of examples is not as important as the
'quality'. The examples need to be selected very carefully
if the system is to perform reliably and efficiently.
Neural Networks are used experimentally to model the
human cardiovascular system.
Diagnosis can be achieved by building a model of the
cardiovascular system of an individual and comparing it
with the real time physiological measurements taken
from the patient.
If this routine is carried out regularly, potential harmful
medical conditions can be detected at an early stage and
thus make the process of combating the disease much
easier
The evaluation sample is a sample of the in-vehicle noise
samples recorded using the simulation head. The sample
samples the noise samples of the front and rear seats of
the car at different fan speeds. A total of 34 samples, each
sample length of 5s.Many studies have shown that
loudness differences have a significant effect on the
evaluation results, and their correlations are about 90%.
Therefore, the loudness of the evaluation samples is
equalized to obtain the correct evaluation results of the
evaluated parameters.
In this experiment, the grouping method is used. The
method is to participate in the evaluation of 34 sound
samples (according to l-34 on its series) To select one of
the sound samples as a reference to the (anchor) sample,
that is, each group contains reference sound samples. In

the case of grouping the samples, the uniform
distribution of the evaluation characteristics of each set
of samples will result in a better consistency of the
evaluation results between the groups.
2.2 The numerical results of the evaluation:
After the evaluation result of the evaluation subject is
obtained, it is digitized. In this experiment, it is assumed
that the comparison of the amount of the sample before
the sample is greater than the subsequent sample, its
assignment is 1; otherwise, the assignment is -1; that the
two are equal when the assignment is 0.The evaluation
of the poor performance of the data were removed,
because A, B, C three groups of samples and evaluation is
relatively independent, so the results of the main body
removed also maintained its relative independence, each
other is not necessarily linked.
2.3 Evaluation results:
After the scores of each group are analyzed, the results of
each group need to be combined to obtain the sorting
results of all the samples. The results of each group can
be combined based on the reference samples designed in
the evaluation experiment.
III. BP NEURAL NETWORK MODEL:
Feedback networks can have signals traveling in both
directions by introducing loops in the network.
A Neural Network can be successfully used in the C.A.I.S.
project, given that we have access to enough training
data.
Since we need it to be highly accurate, we need to
carefully gather contextual variables that closely relate
to diagnosing heart murmur and its various grades.
Feedback networks are dynamic; their 'state' is changing
continuously until they reach an equilibrium point.
They remain at the equilibrium point until the input
changes and a new equilibrium needs to be found.
Feedback architectures are also referred to as interactive
or recurrent, although the latter term is often used to
denote feedback connections in single-layer
organizations.
BP network is a multi-layer feed forward network with
single hidden layer trained by back propagation
algorithm. It is a mature and widely used artificial neural
network. It is a kind of multi-layer forward feedback
neural network with error back propagation .In theory, a
three-tier BP network can approach any nonlinear
function.
The topology of the BP neural network is 4-15-15-1, and
the structure of the network model.The individual
relations between the input variables and the output
variables are not developed by engineering judgment so
that the model tends to be a black box or input/output
table without analytical basis. The sample size has to be
large. Requires lot of trial and error so training can be
time consuming.
A total of 24 data from 1 to 24 in Table 1 were selected
as training samples, and data of 25 to 34 were selected
as test samples. The use of software MatlabR2007 neural

network toolbox (Neural Network Toolbox) newff
function to create BP training network; use trained as a
training function, the function is studied by the gradient
descent method, and the learning rate is adaptive.
It can be seen from Fig.2 that the established BP neural
network prediction model shows good prediction
accuracy and fast convergence speed (after 214 training,
the network error is required).
IV. CONCLUSIONS:
Based on the BP neural network, the acoustic quality
prediction of the vehicle noise can be considered as a
quantitative factor, which can be a quantitative factor or
a qualitative or uncertain factor, which can solve the
nonlinear problem in the acoustic quality prediction. The
BP neural network prediction model, which is
established in this paper, not only provides some
guidance for the prediction of the noise quality in the
vehicle, but also proves the feasibility of the model of
noise prediction in the vehicle noise.
REFERENCES:
[1] Sottek R, Genuit K. Sound quality evaluation of fan
noise based on advanced hearing-related parameters[J].
Noise Control Engineering Journal, 2009, 57(4): págs.
384-390.
[2] Z wicker，E .& Fastl H .Psychoacoustics: Facts and
models.2nd Edition Springer-Verlag Berlin Heidelberg
New York, 1999.
[3] ISO 532-1:2017 Acoustics -- Methods for calculating
loudness -- Part 1: Zwicker method.
[4] ISO 532-1:2017 Acoustics--Methods for calculating
loudness--Part 2: Moore-Glasberg method.
[5] DIN 45631/A1Calculation of loudness level and
loudness from the sound spectrum -Zwicker method. [6]
DIN 45692

IRJET-Sound-Quality Predict for Medium Cooling Fan Noise based on BP Neural Network Model

More Related Content

What's hot (19)

Similar to IRJET-Sound-Quality Predict for Medium Cooling Fan Noise based on BP Neural Network Model (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET-Sound-Quality Predict for Medium Cooling Fan Noise based on BP Neural Network Model