Expert Systems with Applications 41 (2014) 3993–3999

Contents lists available at ScienceDirect

Expert Systems with Applications

journal homepage: www.elsevier.com/locate/eswa

Data mining with various optimization methods

Vladimir Nedic a, Slobodan Cvetanovic b, Danijela Despotovic c, Milan Despotovic d,⇑, Sasa Babic e
a
Faculty of Phil. and Arts, University of Kragujevac, Jovana Cvijica bb, 34000 Kragujevac, Serbia
b
Faculty of Economics, University of Nis, Trg kralja Aleksandra Ujedinitelja 11, 18000 Nis, Serbia
c
Faculty of Economics, University of Kragujevac, Djure Pucara Starog 3, 34000 Kragujevac, Serbia
d
Faculty of Engineering, University of Kragujevac, Sestre Janjic 6, 34000 Kragujevac, Serbia
e
College of Applied Mechanical Engineering, Trstenik, Serbia

a r t i c l e i n f o a b s t r a c t

Keywords: Road traffic represents the main source of noise in urban environments that is proven to significantly
Traffic noise affect human mental and physical health and labour productivity. Thus, in order to control noise sound
Artificial intelligence level in urban areas, it is very important to develop methods for modelling the road traffic noise. As
Genetic algorithm observed in the literature, the models that deal with this issue are mainly based on regression analysis,
Hooke and Jeeves
while other approaches are very rare. In this paper a novel approach for modelling traffic noise that is
Simulated annealing
Particle swarm optimization
based on optimization is presented. Four optimization techniques were used in simulation in this work:
Software genetic algorithms, Hooke and Jeeves algorithm, simulated annealing and particle swarm optimization.
Two different scenarios are presented in this paper. In the first scenario the optimization methods use
the whole measurement dataset to find the most suitable parameters, whereas in the second scenario
optimized parameters were found using only some of the measurement data, while the rest of the data
was used to evaluate the predictive capabilities of the model. The goodness of the model is evaluated by
the coefficient of determination and other statistical parameters, and results show agreement of high
extent between measured data and calculated values in both scenarios. In addition, the model was com-
pared with classical statistical model, and superior capabilities of proposed model were demonstrated.
The simulations were done using the originally developed user friendly software package.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction traffic management, which is particularly important for human

Road traffic noise along with the noise coming from railways
and industries represents very important factor regarding environ-
mental pollution in urban areas. The influence of traffic noise on
human health has been studied on numerous occasions in recent
years (Brink, 2011; Fyhri & Klboe, 2009; Pirrera, De Valck, &
Cluydts, 2010) resulting that this kind of annoyance significantly
affects both mental and physical health in many ways: causing
anxiety, stress, hearing impediments, sleep disturbance, cardiovas-
cular problems, etc. Thus, in order to control noise sound level in
urban areas, it is very important to develop methods for prediction
of the traffic noise. Due to the rapid development of means of
transportation and road traffic, the influence of the traffic flow
structure on the level of traffic noise is an important area of
research. Through the monitoring of basic flow parameters and
their trends it is possible to predict and monitor noise that appears
in the certain part of the transport network. In this way, the effect
of noise reduction can be achieved through different modes of
⇑ Corresponding author. Tel.: +381 69 844 9679.
E-mail address: mdespotovic@kg.ac.rs (M. Despotovic).
health and environmental improvement.
The first traffic noise prediction (TNP) models date back to early
1950s. Since then large number of methods and models for traffic
noise prediction has been developed. The critical reviews of the
most used ones are given in Steele (2001) and Quartieri et al.
(2009). Most of the TNP models that are presented in literature
are based on linear regression analysis. The main limit of those
models, as concluded in Quartieri et al. (2009) and Guarnaccia,
Lenza, Mastorakis, and Quartieri (2011), is ‘‘that they do not take
into account the intrinsic random nature of traffic flow, in the
sense that they do not take care of how vehicles really run, consid-
ering only how many they are’’. More advanced models involve
artificial neural networks (ANN) (Cammarata, Cavalieri, & Fichera,
1995; Givargis & Karimi, 2010) and genetic algorithms (Gndogdu,
Gkdad, & Yksel, 2005; Rahmani, Mousavi, & Kamali, 2011). ANN
model that was used in Cammarata et al. (1995) has 3 inputs:
equivalent number of vehicles, which was obtained by adding to
the number of cars number of motorcycles multiplied by 3 and
number of trucks multiplied by 6, the average height of the build-
ings on the sides of the road, and the width of the road. In order to
increase the number of inputs authors decomposed equivalent
number of vehicles into the number of cars, the number of

0957-4174/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.eswa.2013.12.025
3994 V. Nedic et al. / Expert Systems with Applications 41 (2014) 3993–3999

motorcycles, and the number of trucks, and got the ANN model
with 5 inputs. In terms of the parameters involved in the CoRTN
(Calculation of Road Traffic Noise) model (Quartieri et al., 2009),
which was initially developed in 1975 by the Transport and Road
Research Laboratory and the Department of Transport of the Uni-
ted Kingdom, the ANN model that was used in Givargis and Karimi
(2010) has 5 input variables: the total hourly traffic flow, the
percentage of heavy vehicles, the hourly mean traffic speed,
the gradient of the road, and the angle of view. Authors tested
the developed model on the data collected on Tehran’s roads,
and found no significant differences between the outputs of the
developed ANN and the calibrated CoRTN model. In Gndogdu
et al. (2005) genetic algorithm was used to model the traffic noise
in relation to traffic composition (vehicle per hour), the road gradi-
ent and the ratio of building height to the road width. In Rahmani
et al. (2011) the proposed model is a function of total equivalent
traffic flow and equivalent traffic speed. In both papers the authors
used MATLAB to find the optimized values of model parameters.
In this paper an application of four optimization techniques for
the prediction of traffic noise is presented. These techniques are:
genetic algorithms, Hooke and Jeeves algorithm, simulated anneal-
ing, and particle swarm optimization. The model that is proposed
consists of five variables: the number of light motor vehicles, the
number of medium trucks, the number of heavy trucks, the num-
ber of buses and the average traffic flow speed. All optimized mod-
els are tested on data measured on Serbian road using the
originally developed user friendly software package.
2. Problem formulation
The most suitable measure for depicting traffic noise emission
is equivalent sound pressure level ðLeq Þ, which is expressed in units
of dbA and corresponds to fictitious noise source emitting steady
noise, which in specific period of time contains the same acoustic
energy as the observed source with fluctuating noise. For a number
of discrete measurements ðNÞ; Leq for time period T is expressed by
following equation:
!
X
N Li
Leq ¼ 10log 10 1=T 1010
i¼1
possible. Simultaneously, variations in traffic flow, traffic speed
and composition of traffic flow were measured. For that reasons
the surveys at the same time also consist of the following param-
eters: the number of light motor vehicles, the number of medium
trucks, the number of heavy trucks, the number of buses, and the
average traffic speed in the given time periods.
Measurements were taken in accordance with recommenda-
tions for road traffic noise measurement; microphone was
mounted away from reflecting facades, at a height of 1.2 m above
the ground level and 7.5 m away from central line of the road. Dur-
ing the measurements it has been taken care that climate condi-
tions are as similar as possible (no wind, no rain) in order to
eliminate their influence.
4. Mathematical model and methods
The equivalent sound pressure level is supposed to be modeled
by the following equation:
Leq ¼ N1  log 10 ðLMVÞ þ N2  log 10 ðSTVÞ þ N3  log 10 ðTTVÞ
þ N4  log 10 ðBUSÞ þ N5  Vav g N6 þ N7  log 10 ðVav gÞ ð2Þ
where Ni ði ¼ 1  7Þ are coefficients. The problem transforms to find
coefficients Ni , such that supposed model best fits experimental data.
For that purpose genetic algorithms, Hooke and Jeeves algorithm,
simulated annealing, and particle swarm optimization are used.
These techniques are briefly described in following subchapters.
4.1. Genetic algorithms
Genetic algorithms (Rao, 1996) are class of evolutionary algo-
rithms that could be used for a large number of different applica-
tion areas. The principle of genetic algorithms is based on
Darwin’s theory of evolution, by which the fittest individuals have
the best chances to survive. Genetic algorithms operate with a set
of individuals (chromosomes) called population. The information
ð1Þ

th
where Li is sound pressure level, which corresponds to i
measurement.
In order to reduce the noise it is necessary to know functional
relationship between the equivalent sound pressure level and
influential parameters. Leq is correlated to numerous parameters,
such as numbers and types of vehicles, their velocities, type of road
surface, width and slope of the road, height of buildings facing the
road, etc. As mentioned in the introduction, in this paper the
following variables were considered: the number of light motor
vehicles (LMV), the number of medium trucks (STV), the number
of heavy trucks (TTV), the number of buses (BUS) and the average
traffic flow speed (Vavg). A brief description of how these variables
were measured is given in the following chapter.

3. Data sampling

For traffic data measurement and for noise measurement on the

road M5, automatic traffic counters QLTC-10C and sound level me-
ter Bruel&Kajer type 2230 class 1 respectively were used. The
equivalent sound pressure levels were measured for time period
of 15 min. In order to include greater number of scenarios that
might occur in urban environments, a total of 124 measurements
of equivalent noise levels for time periods of 15 min were carried
out. Measurements of Leq for time period of 15 min were performed
at various times to include diversity of the traffic flow as much as Fig. 1. Flowchart of the Genetic algorithm workflow.
 V. Nedic et al. / Expert Systems with Applications 41 (2014) 3993–3999
that is encoded in each chromosome represents possible solution
to an optimization problem.
A GA workflow that is used for optimization is shown on Fig. 1.
In this work fitness is calculated as average difference between
measured (target) value and calculated value of the given model on
a whole dataset. In this context, the lower the fitness, the better the
chromosome is.
4.2. Hooke and Jeeves
The Hooke and Jeeves pattern search algorithm (Rao, 1996) is a
derivative free optimization algorithm that can be used to optimize
functions which are not continuous or differentiable. Hooke and
Jeeves algorithm consists of a sequence of exploratory moves about
a base point which, if successful, are followed by pattern moves.
During the exploratory search, the algorithm gathers information
about the function to be optimized in the neighborhood of the cur-
rent base point by searching the values of the function in direction
of all decision variables at distance equals step size. If exploratory
search cannot improve the solution in any direction the step size
used in exploratory search is reduced and the search is repeated
at the previous base point. If there is a reduction in function value
then the new value is retained and after all the variables have been
considered, a new base point will be defined. The algorithm termi-
nates when either the maximum number of iterations is reached or
the step size becomes smaller than a predefined stopping criterion.
4.3. Simulated annealing
The simulated annealing algorithm (Rao, 1996) was originally
inspired from the process of annealing in metallurgy, in which
metals are heated up to high temperature and then cooled very
slowly. Slowly cooling allows the metal to alter its physical proper-
ties and achieve its most regular crystal lattice configuration which
corresponds to minimal energy state, which might not occur if me-
tal is cooled too fast. Simulated annealing involves temperature
variable, which simulates this heating process. This variable is ini-
tially set to high value and then gradually diminishes as the algo-
rithm runs. Temperature reduction is specified with annealing
schedule, which most often is defined as geometric cooling. Geo-
metric cooling means that temperature in each step is multiplied
by temperature reduction factor, which is less than 1. The key algo-
rithmic feature of simulated annealing is its ability to escape local
optima by allowing changes that worsen the objective function va-
lue according to its probability function. The probability of accept-
ing non-improving solutions depends on a temperature parameter,
which is reduced over time; hence when the temperature variable
is high the algorithm allows accepting worse solutions more fre-
quently. This makes it possible that in early stage of execution
algorithm jumps out of any local optimums. As the temperature
is reduced the chance of accepting worse solutions is also reduced.
The simulated annealing algorithm is presented as follows:
Step 1: Randomly choose an initial solution x0 . Begin with an ini-
tial temperature T ¼ T 0 and calculate the function value at
initial solution fold .
Step 2: Create a new solution xn by adding a small perturbation to
previous solution. Calculate the function value at new
solution fnew .
Step 3: If the new solution is better ðD ¼ fnew  fold < 0Þ, then it is
accepted; else accept the new solution only if RND (0, 1)
< exp (-D/T).
Step 4: Repeat steps 2 and 3 certain number of iterations.
Step 5: Reduce temperature (set T = temperature reduction factor
⁄ T) and repeat algorithm starting from step 2 until tem-
perature is lower than some predefined value.
3995
4.4. Particle swarm optimization
Particle Swarm Optimization (PSO) is stochastic optimization
technique invented by Russell Eberhart and James Kennedy in
1995 (Kennedy & Eberhart, 1995). PSO is inspired by collaborative
behaviour of biological populations such as flocks of birds or
schools of fish. In the PSO algorithm each individual in a swarm
is called a ‘‘particle’’, and has a fitness value, which is evaluated
by the fitness function to be optimized and represents how good
it solves the problem. In addition, each particle has a velocity that
directs the iterative movement (flying) of the particle in a multidi-
mensional search space. It is affected by particle’s inertia, the loca-
tion of the best fitness achieved so far by the particle itself (particle
best) and by the location of the best fitness achieved so far in the
whole swarm (global best). In every iteration ðkÞ of PSO algorithm,
each particle ðiÞ updates its velocity ðv i Þ and position ðxi Þ by the fol-
lowing two equations:
   
k k
v kþ1
i ¼ xv ki þ c1 r k1;i Pbesti  xki þ c2 rk2;i Gbest  xki ð3Þ
xkþ1
i ¼ xki þ v kþ1
i ð4Þ
where x is an inertia weight, c1 and c2 are acceleration coefficients,
k
rk1;i and r k2;i are random numbers in the interval [0, 1], Pbesti and
k
Gbest are best positions achieved so far by the particle itself and
the whole swarm respectively.
Mechanism of a PSO technique using the velocity and the posi-
tion update rules is illustrated in Fig. 2.
The PSO algorithm can be summarized as follows:
1. Initialize each particle with a random velocity and random
position.
2. Calculate fitness values for each particle; if the current fit-
ness is better than the particle best value so far, save current
position as particle best.
3. Choose the particle with the best fitness of all particles and
assign its fitness value to global best.
4. Calculate, for each particle, the new velocity and position
according to update rules.
5. Repeat steps 2–4 until an interruption criterion is reached.
5. Simulation
The simulation was done using the home made originally de-
signed user friendly software (Fig. 3).
To assess the quality of proposed models two scenarios were
applied. In the first scenario the whole measurement dataset was
used to find the model that best fits those data, whereas in the sec-
ond scenario we wanted to test also the predictive capabilities of
Fig. 2. Particle position and velocity updates.
3996 V. Nedic et al. / Expert Systems with Applications 41 (2014) 3993–3999

Fig. 3. GUI of the software that was used for simulation.

developed model. Therefore the measurement dataset was split in
two parts by randomly selecting data. The first part, which con-
tains 100 data (80% of total number of measurements), was used
to optimize the model, and the second part, which contains other
24 data (20% of total number of measurements), was used to test
the predictive capabilities. In both scenarios the goodness of fit is
assessed by the coefficient of determination ðR2 Þ and other statis-
tical parameters.
The parameters that were used for each optimization method
are given in Table 1.
coefficient of determination ðR2 Þ. A comparative statistical tests
(F-test and Paired t-test) were applied to compare calculated traffic
noise levels with the measured values.
These statistical analysis and statistical tests are applied to var-
ious optimization methods that were used to find the optimum
model parameters. In addition, in order to compare the proposed
model with classical regression models for traffic noise modelling,
the same analysis and tests are applied to widely used statistical
regression model developed by Fagotti et al. (Quartieri et al.,
2009). The Fagotti regression model is given by following
expression:

6. Results Leq ¼ 10logðQ L þ Q M þ 8Q P þ 88Q BUS Þ þ 33:5 ð5Þ

The optimized parameters of the proposed model for various
optimization methods, for scenario 1 are shown in Table 2. The
optimized parameters of the proposed model for various optimiza-
tion methods, for scenario 2 are shown in Table 3.
Capabilities of the models to calculate noise levels are assessed
by certain statistical parameters and by performing statistical
tests. The goodness of fit of the models are appraised using the
mean error (ME), the mean absolute error (MAE), the mean
absolute relative error (MARE), coefficient of correlation ðRÞ and
where Q L is the number of light vehicles per hour, Q P is the number
of heavy vehicles per hour, Q M is the number of motorcycles per
hour, and Q BUS is the number of buses per hour.
The comparative analysis for scenario 1 is presented in Table 4,
whereas the comparative analysis for scenario 2 (testing the model
with the data that was not used in optimization) is presented in
Table 5. Based on these data it could be concluded that there is
relatively good agreement between measured data and computa-
tional results for all optimization methods for both scenarios. This

Table 1
Parameters that were used for optimization.

GA Hooke & Jeeves Simulated annealing PSO

Population size: 100 Step size Initial temperature: 10 Number of particles: 100
Selection method: Roulete wheeel reduction factor: 0.5 Temp. reduction factor: 0.99 Inertia weight: 0.5
Crossover: Uniform Final temperature: 10 5 c1 : 0.5
Crossover rate: 0.7 c2 : 0.5
Substitution: Weakest
 V. Nedic et al. / Expert Systems with Applications 41 (2014) 3993–3999 3997

Table 2
Optimized parameters found by various optimization methods for whole dataset (NDATA = 124).

Opt. method N1 N2 N3 N4 N5 N6 N7
GA 3.8 0.322 3.729 1.114 14.464 0.245 29.997
Hooke & Jeeves 3.166 0.358 3.726 0.908 20.51 0 22.36
Sim. Annealing 3.603 0.555 3.567 1.158 155.662 182.254 33.087
PSO 2.799 0.703 3.77 1.118 98.684 0.916 32.671

Table 3
Optimized parameters found for modelling dataset (NDATA = 100) for various optimization methods.

Opt. method N1 N2 N3 N4 N5 N6 N7
GA 6.112 0.301 3.29 0.961 29.918 0.857 29.96
Hooke & Jeeves 2.741 0.949 3.391 0.994 13.742 0.304 7.085
Sim. Annealing 3.49 0.831 3.19 1.123 46.183 113.006 33.301
PSO 3.327 0.73 3.312 1.189 58.232 0.788 32.263

Table 4
Statistical analysis for various optimization methods for whole dataset (NDATA = 124, Scenario 1).

R R2 ME MAE MARE F value t value

GA 0.949 0.900 0.052 0.307 0.004 1.060 0.335

Hooke & Jeeves 0.946 0.895 0.013 0.315 0.004 1.249 0.087
Sim. Annealing 0.947 0.897 0.030 0.317 0.004 1.025 0.188
PSO 0.948 0.899 0.077 0.304 0.004 1.071 0.500
Fagotti et al. 0.629 0.395 11.512 11.512 0.193 1.014 73.490

tka  table ¼ 1:96ða ¼ 5%; k ¼ 246Þ

F ka1 ;k2  table ¼ 1:21ða ¼ 5%; k1 ¼ 123; k2 ¼ 123Þ

Table 5
Statistical analysis for various optimization methods for test dataset (NDATA = 24, Scenario 2).

R R2 ME MAE MARE F value t value

GA 0.95 0.903 0.048 0.328 0.005 1.166 0.131

Hooke & Jeeves 0.954 0.911 0.183 0.357 0.005 1.017 0.475
Sim. Annealing 0.969 0.938 0.047 0.28 0.004 1.199 0.128
PSO 0.975 0.951 0.026 0.269 0.004 1.306 0.073
Fagotti et al. 0.555 0.308 11.538 11.538 0.192 1.275 31.949

t ka  table ¼ 1:96ða ¼ 5%; k ¼ 46Þ

F ka1 ;k2  table ¼ 1:44ða ¼ 5%; k1 ¼ 47; k2 ¼ 47Þ

Fig. 4. Side by side comparison of measured data and calculation obtained using GA Fig. 5. Comparison of measured data (T) and calculation (A) obtained using GA for
for scenario 1. scenario 1.
3998 V. Nedic et al. / Expert Systems with Applications 41 (2014) 3993–3999
Fig. 6. Predictive capabilities of proposed models that are obtained using different optimization methods.
is also obvious from Fig. 4 that shows side by side comparison of
measured data and calculation values and Fig. 5 that shows those
data in normalized coordinate system. For the sake of conciseness
only results obtained using GA are presented.
The correlation coefficient between the measurements and
calculated values of proposed model is relatively high, which indi-
cates that there is a close relationship between the measurements
and the results of model for all optimization methods. On the other
hand Fagotti regression shows by far worse results (R ¼ 0:629 for
scenario 1 and R ¼ 0:555 for scenario 2).
The high values of coefficient of determination for both scenar-
ios imply good modelling and predictive capabilities of proposed
model. All statistical parameters show that the proposed model
is superior to classical regression model, particularly concerning
predictive capabilities (Table 5).
The null hypothesis of the F test shows that the dispersion of
the calculated and measured noise values are approximately equal.
For the degrees of freedom k1 ¼ 23; k2 ¼ 23 and p < 0:05 calcu-
lated F value for various optimization methods and Fagotti model
is less than the table value of F 23;23
0:05 ¼ 1:44 (Table 5). It can be con-
cluded that the dispersion of the measured and calculated values
do not differ significantly, and there is no basis for rejecting the
null hypothesis of equal dispersion of data sets. When the paired
t-test is concerned the mean value of the measured and calculated
noise values are approximately equal. For the degree of freedom
k ¼ 46 and p < 0:05 calculated t is less than the table value
t46
0:05 ¼ 1:96 (Table 5), and there is no reason to reject the null
hypothesis of equality of means of data sets. This is not case with
Fagotti model where calculated t value is more than the table value
and the null hypothesis of equality of means of data sets must be
rejected.
Predictive capabilities are illustrated in Fig. 6, where side by
side comparison of measured data that was not used in optimiza-
tion, calculated values obtained using various optimization meth-
ods and calculated values obtained using Fagotti model is shown.
7. Conclusion
In this paper a novel approach for modelling of traffic noise is
presented. This approach is based on advanced optimization tech-
niques on the contrary to regression analysis that is most widely
used method for traffic noise modelling. Four optimization
techniques were used to optimize the proposed model: genetic
algorithms, Hooke and Jeeves algorithm, simulated annealing,
and particle swarm optimization. The model is optimized and
tested using originally developed user friendly software package.
The goodness of the model is evaluated by statistical parameters
and compared with experimental results and widely used regres-
sion model for traffic noise modelling developed by Fagotti et al.
All statistical analysis show that the developed model is precise.
Statistical tests showed a high correlation of proposed model to
the measured values which makes the model significant.
The proposed model operates with relatively simple input val-
ues (parameters and structure of traffic flow) that are easily mea-
surable on site, and that are already monitored on the traffic
network. Validation has shown that by measuring and monitoring
traffic flow parameters it is possible by means of developed model
to predict traffic noise level by far better than with classical statis-
tical models. That makes it possible to take certain measures
toward the traffic noise reducing by changing the influential traffic
flow parameters and traffic regime. Based on calculated noise level
by means of proposed model, it is possible to appraise the noise
level in existing as well as in newly designed traffic arteries or de-
tours, or during the traffic regime change on an existing network.
Traffic noise is complex phenomenon that depends on many
influential factors. Apart from intrinsic parameters such as
numbers of vehicles in specific group of vehicles and the average
speed of the traffic flow, there is also a group of parameters that
are specific for area under investigation, such as speed limits, road
surface, driving skills and habits, type of intersections, traffic lights’
schedules etc. The proposed model for the analysis of traffic noise
is shown to be a reliable tool for the practical application of the
calculating equivalent noise levels based on the traffic flow struc-
ture for typical road in Serbia. The weakness of the model is that
it does not take into account the area specific parameters. There-
fore the future research directions might be to include these
specific parameters in model in order to make it more general.
References
Brink, M. (2011). Parameters of well-being and subjective health and their
relationship with residential traffic noise exposurea representative evaluation
in switzerland. Environment International, 37, 723–733.
Cammarata, G., Cavalieri, S., & Fichera, A. (1995). A neural network architecture for
noise prediction. Neural Networks, 8, 963–973.
 V. Nedic et al. / Expert Systems with Applications 41 (2014) 3993–3999
Guarnaccia, C., Lenza, T. L. L., Mastorakis, N. E., & Quartieri, J. (2011). A comparison
between traffic noise experimental data and predictive models results.
International Journal of Mechanics, 5, 379–386.
Fyhri, A., & Klboe, R. (2009). Road traffic noise, sensitivity, annoyance and self-
reported healtha structural equation model exercise. Environment International,
35, 91–97.
Givargis, S., & Karimi, H. (2010). A basic neural traffic noise prediction model for
tehrans roads. Journal of Environmental Management, 91, 2529–2534.
Gndogdu, O., Gkdad, M., & Yksel, F. (2005). A traffic noise prediction method based
on vehicle composition using genetic algorithms. Applied Acoustics, 66, 799–809.
Quartieri, J., Mastorakis, N. E., Iannone, G., Guarnaccia, C., D’Ambrosio, S., Troisi, A., &
Lenza, T. L. L. (2009). A review of traffic noise predictive models. In The 5th
3999
WSEAS international conference on applied and theoretical mechanics, Puerto De
La Cruz, Canary Islands.
Kennedy, J., & Eberhart, R. (1995). Particle swarm optimization. In Proc. of IEEE
international conference on neural network, Piscataway, NJ (pp. 1942–1948).
Pirrera, S., De Valck, E., & Cluydts, R. (2010). Nocturnal road traffic noise: A review
on its assessment and consequences on sleep and health. Environment
International, 36, 492–498.
Rahmani, S., Mousavi, S. M., & Kamali, M. J. (2011). Modeling of road-traffic noise
with the use of genetic algorithm. Applied Soft Computing, 11, 1008–1013.
Rao, S. S. (1996). Engineering optimization: Theory and practice. John Wiley & Sons.
Steele, C. (2001). A critical review of some traffic noise prediction models. Applied
Acoustics, 62, 271–287.