Li, Daming; Deng, Lianbing; Su, Qinglang; Song, Yong-Hua

Article
Providing a guaranteed power for the BTS in telecom
tower based on improved balanced owl search algorithm

Energy Reports

Provided in Cooperation with:

Elsevier

Suggested Citation: Li, Daming; Deng, Lianbing; Su, Qinglang; Song, Yong-Hua (2020) : Providing a
guaranteed power for the BTS in telecom tower based on improved balanced owl search algorithm,
Energy Reports, ISSN 2352-4847, Elsevier, Amsterdam, Vol. 6, pp. 297-307,
https://doi.org/10.1016/j.egyr.2020.01.006

This Version is available at:

https://hdl.handle.net/10419/244032

Standard-Nutzungsbedingungen: Terms of use:

Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Documents in EconStor may be saved and copied for your personal
Zwecken und zum Privatgebrauch gespeichert und kopiert werden. and scholarly purposes.

Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle You are not to copy documents for public or commercial purposes, to
Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich exhibit the documents publicly, to make them publicly available on the
machen, vertreiben oder anderweitig nutzen. internet, or to distribute or otherwise use the documents in public.

Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen If the documents have been made available under an Open Content
(insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, Licence (especially Creative Commons Licences), you may exercise
gelten abweichend von diesen Nutzungsbedingungen die in der dort further usage rights as specified in the indicated licence.
genannten Lizenz gewährten Nutzungsrechte.

https://creativecommons.org/licenses/by-nc-nd/4.0/
 Energy Reports 6 (2020) 297–307

Contents lists available at ScienceDirect

Energy Reports
journal homepage: www.elsevier.com/locate/egyr

Research paper

Providing a guaranteed power for the BTS in telecom tower based on

improved balanced owl search algorithm
∗
Daming Li a,b,c , Lianbing Deng a,b,d , , Qinglang Su c , Yonghua Song e
a
The Post-Doctoral Research Center of Zhuhai Da Hengqin Science and Technology Development Co., Ltd, Guangdong Hengqin New
Area, 519031, China
b
Guangdong Qinzhi Science and Technology Research Institute, Guangdong Hengqin New Area, 519031, China
c
Institute of Data Science, City University of Macau, Macau, 999078, China
d
Huazhong University of Science and Technology, Hubei Wuhan, 430074, China
e
Macao University, Macau, 999078, China

article info a b s t r a c t

Article history: The telecommunication industry is the technology of communication among humankind. One problem
Received 13 October 2019 for this industry is to provide the required energy for its towers in the remote rural areas in the absence
Received in revised form 27 November 2019 of the grid. This objection causes dependency of this industry to the diesel generators and battery banks
Accepted 21 January 2020
as backup resources. Despite not being resilient and the effect of environment for the power supplies in
Available online xxxx
diesel generators is still widespread. This paper presents a new optimized and environmentalist power
Keywords: supply configuration for the telecommunication tower based on the proton exchange membrane fuel
Telecom tower cell (PEMFC). An optimized PI-based converter is also utilized to keep the source value reliable. For
PEMFC achieving optimized results for the PI controller, a developed version of the owl search algorithm is
Owl search algorithm proposed. Simulation results have been applied in different conditions and the results are compared
Improved
with Ziegler–Nichols, particle swarm optimization algorithm, and emperor penguins optimizer based
Boost converter
PI controller to show the excellence of the proposed method.
DC distribution
Voltage regulation © 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).

1. Introduction mobile networks. The antennas installed on these towers are

connected to the network via microwave radio waves, trying to
The world of telecommunication and information production get a city-like area under the full coverage of a mobile network.
is rapidly changing, and today we are witnessing their conver- Hence, providing a robust and non-stop power supply guarantees
gence with each other so that data and information are quickly customer service quality (Kaur et al., 2017). In distant areas, guar-
and unimaginably transmitted to users around the world (Choi anteed power supplying for BTSs is not an easy task, especially
et al., 2019). It has undoubtedly led to widespread developments due to the unavailability of the national grid in this area.
in all the social and economic spheres of humanity and its im- Finding a reliable and economical solution to supply the BTS
pact on human societies in a way that today’s world is rapidly power installed in rural areas and away from the grid has be-
becoming a society based on telecommunication. The scope of its come a challenge for mobile operators. Therefore, reviewing the
use and impact on different aspects of today’s and future life of available options for the BTS power supply in the telecommuni-
human societies has become one of the most important topics in cations industry is crucial. One of the best appropriate selections
the world today and has attracted the attention of many countries for supplying a non-stoppable power for the distant areas is to
around the world (Mahajan et al., 2019). A significant issue which use diesel generators (DG) (Ojo et al., 2019). A DG is a small
is recently mentioned in the telecommunication industry is the power plant that usually takes on the task of generating electric-
supply problem for the base transceiver station (BTS) of telecom- ity during an emergency that cuts off power for any reason. It
munication tower, especially for the distant rural zones. BTS is can also as a fixed generator for the distant areas. A significant
a set of wireless and wired telecommunications equipment that shortcoming of the DGs is that they have a high startup failure
facilitates communication between a mobile user and a public which is about 15%. To solve this problem, a battery bank can
network. BTS stations are the most important component of GSM be employed (Paudel and Wasti, 2018). Since the environmental
hazard impact of DGs, organizations are working on upgrading
∗ Corresponding author at: Guangdong Qinzhi Science and Technology these power supplies collection based on green technologies like
Research Institute, Guangdong Hengqin New Area, 519031, China. renewable technologies. Recently, the use of renewable energy
E-mail address: [email protected] (L. Deng). technology such as wind systems, fuel cells, and photovoltaic has

https://doi.org/10.1016/j.egyr.2020.01.006
2352-4847/© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
298 D. Li, L. Deng, Q. Su et al. / Energy Reports 6 (2020) 297–307

been increased dramatically due to their easy access, the unend- Plus, due to the absence of mechanical parts like rotating parts in
ing and the non-pollution of these sources to supply electricity to PEMFCs, they have not mechanical wear which makes them more
off-grid areas. reliable.
This makes the application of renewable energy resources PEMFCs require hydrogen for feeding; this can be supplied
important in the telecommunication industry (Hallur and Sane, by the bio-fuels or gas. This can be supplied by the nearby gas
2018) such that the application of renewable energy resources pipelines. For feeding the PEMFC, hydrogen is made by elec-
like photovoltaic energy conversion systems and wind energy has trolyzing operation powered by different resources (Gheydi et al.,
been developed as the backup power supplies of the DGs (Agha- 2016). From the literature, it can be concluded that the com-
jani and Ghadimi, 2018; Liu et al., 2017; Gollou and Ghadimi, bination of the PEMFC with the BTS in the telecom tower can
2017). In recent years, among different types of renewable re- guarantee an adjustable energy resource for the BTS distribu-
sources, fuel cells are turning into one of the more interesting tion bus in different operations. The voltage of PEMFC can be
technologies for the long term backup for the telecommunication changed by oscillations in the voltage and temperature. This
industry (Mirzapour et al., 2019). This is due to the fuel cell’s makes unreliable results for supplying the BTS station as a backup
characteristics such as higher efficiency for conversion, longer system.
runtime (greater than 8 h), quiet operation, lower pollutant to the In this study, a new optimized boost converter interfacing unit
environment, low cost of maintenance, and rooftop installation is proposed for covering this drawback. A boost converter is a
capability. component that levels up the voltage to a considered voltage level
There are plenty of research works about this category. For in DC bus. The research provides a new modified optimization
instance, Hosseini Firouz and Ghadimi (2016) introduced a robust algorithm, called Improved Owl search algorithm, for optimal
configuration for a power system including the fuel cell as backup control of the boost component for different reasons such as
supply in the telecommunication applications. The method was providing a fast response for the dynamic of the interfacing
testified in the practical position in a station with GSM and unit, adjusting the DC distribution voltage, and guaranteeing the
the results give 98.5% robustness for the fuel cell system during reliability of the system under output oscillations of the PEMFC
260 cycles. Hamian et al. (2018) analyzed the effects of fuel cell during load variation of telecom tower. Generally, The novelty
technology along with photovoltaic system for a distant rural and contributions of propose paper is summarized as follow:
telecom system in Bangladesh. Proton exchange membrane fuel
cell (PEMFC) is used as a popular type of fuel cell and then it – Proposing a new optimization algorithm for optimizing the
connected to an upgraded photovoltaic system (solar). The paper PI controller on the boost converter.
claimed that the performance of the PEMFC was increased to – Improving the convergence and exploration shortcomings
57.26%. The lifetime of the fuel cell was also analyzed. of the Owl Search Algorithm based on Chaos theory and LV
Kaur et al. (2017) proposed another method based on proton mechanism.
exchange membrane fuel cells for using in the telecom sup- – Proposing a new clean configuration based on the PEMFC
ply system. The method used a simple closed-loop and genetic for the Telecom.
algorithm for intelligent interfacing unit that contains power – Analyzing the proposed optimized method based on differ-
electronic boost converter for power conditioning. The method ent voltage and load fluctuating.
guaranteed the tight voltage adjustment at the DC distribution
bus of the base transceiver station. The method is simulated in 2. The architecture of the supply system for the BTS station
MATLAB/Simulink platform and verified theoretically. The results
give successful achievements that showed the feasibility of the Fig. 1 shows the main configuration of a BTS station for the
presented system for robust operation of the telecom towers. telecom tower. As it is observed, the BTS distribution system uses
Leng et al. (2018) proposed an on-site hydrogen generation for 48 V DC based on the high DC voltage of the system and its
BTS application for the fuel cell. They used two case studies for security. The component POL (point of loads) converter is adopted
the analysis of the proposed system and the results showed that for providing the load end electronic equipment energy through
the hydrogen can improve the economics of power supply. Mirza- the DC distribution bus.
pour et al. (2019) proposed a comparative analysis for power Switch Mode Power Converter (SMPC) is used here as an
sources of the BTS. The main purpose of the method is to compare electronic power supply to incorporate a switching regulator to
the feasibility of adopting renewable power sources for minimiz- convert electrical power efficiently in the distribution bus of
ing carbon emission, reducing the dependency on fossil fuel, and the base transceiver station. Indeed, the reason for using SMPC
minimizing the cost of the system. A new model is presented to in the system is to provide a filter for the input source and
combine the sources with backup of diesel generator. the telecom load oscillations to obtain a compressed adjusted
Akbary et al. (2019) modeled a hybrid system for powering voltage at the DC bus. This is because of the sensitivity of the
remote BTS. The study’s purpose was about providing a renew- telecommunication load to the voltage oscillations.
able energy source for supplying a remote BTS station. A hybrid In the following, the main architecture of the BTS feeding
renewable energy plant including a wind turbine, photovoltaic system for the present study has been represented. To make a
panels, and a fuel cell was provided. HOMER software was utilized reliable system based on the explanations in the introduction, a
for simulating the system. The final results showed that using 2.5 kW PEMFC is utilized as a clean energy resource to supply
renewable energy sources is feasible solutions to reduce the air the BTS energy by receiving the oxygen and the hydrogen from
contaminants and pollutants generated by sources such as diesel a side to electricity generation based on chemical reaction and
generators. forms water vapor as a by-product. Fig. 2 shows this architecture.
As it is clear from the literature, using renewable energies A significant profit of the PEMFC is that it does not require
for the BTS is exponentially increasing and this is due to its continuous maintenance checking.
advantages that were discussed before (Ebrahimian et al., 2018). As aforementioned, the output voltage of a PEMFC depends
Another case is that in most of these works, PEMFC is used as a on variety of cases from load variations to an environmental
popular fuel cell (Khodaei et al., 2018). This is due to its need to condition like atmospheric pressure, humidity, and temperature
lower operating temperature which makes fewer need to thermal which is changed between 15% to 30% below or above the output
stress on components and warming up time (Bagal et al., 2018). voltage and uses the other parts to keep the outpour voltage in 48
 D. Li, L. Deng, Q. Su et al. / Energy Reports 6 (2020) 297–307 299

Fig. 1. The main configuration of a BTS station for the telecom tower.

Fig. 2. The architecture of the proposed optimized BTS feeding system based on PEMFC.

Table 1 layer of the polymer membrane, producing electricity and water.

The required power for the telecom tower location in different BTS architectures It also provides sensors with the capability of generating power
(Kaur et al., 2017).
in the range of several watts to several megawatts, for use in a
Architecture Load Total power (kW)
variety of applications, including propulsion for ships and trains
DC (kW) AC (kW) to power generation. A PEMFC is a converter to turn the chemical
2 × 2 × 2 1.24 1.8 3.04 reaction between oxygen and hydrogen into electrical energy
4 × 4 × 4 2.04 2.4 4.44 and water. To model this component based on thermodynamic
6 × 6 × 6 2.82 3.6 6.42
principles, the energy available across the cells is achieved as
follows (Firouz and Ghadimi, 2016):
Ea = 1.23 − 8.5 × 10−3 × θfc − 298.15 + 4.31 × 10−5 × T
( )
V at the DC bus. Table 1 illustrates the total power requirement ( √ )
of the telecom tower system for different BTS structures × ln pH2 pO2 (1)
In this study, the DC/DC boost converter in the regulator has
where, θfc represents the fuel cell temperature (Kelvin) and the
been optimized based on a new technique to adjust the output
partial pressure of the oxygen and the hydrogen are as follows:
voltage of the converter. The main purpose of the proposed ⎛ ⎞
optimized regulator is to keep the voltage of the PEMFC output
voltage in the 48 V for the BTS distribution bus. The proposed 1
pO2 = 0.5 × RHa × pst
H2 O × ⎝ ) − 1⎠ (2)
⎜ ⎟
regulator gives also a high-speed response that increases the RHa ×pst
(
H2 O 1.6×(i/A)
pa
exp θ 1.3
speed of the total system.
⎛ ⎞
3. Mathematical model of a fuel cell 1
pH2 = 0.5 × RHa × pst
H2 O × ⎝ ) − 1⎠ (3)
⎜ ⎟
RHc ×pst
(
H2 O 4.2×(i/A)
Generally, PEMFCs are one of the most important sources of pc
exp θ 1.3
clean energy in the coming years. On the other hand, the possi-
And the saturation pressure of the water vapor can be achieved
bility of producing hydrogen from renewable energy sources has
by the following equation:
doubled its attention in recent years. Fuel cells are simple to oper-
H2 O = −2.2 + 0.03 × θfc − 273.15 − 0.92 × 10
log10 pst −4
( ) ( )
ate, with hydrogen and air (or oxygen) as reactants in the catalytic
300 D. Li, L. Deng, Q. Su et al. / Energy Reports 6 (2020) 297–307
)2
× θfc − 273.15
(
Table 2
)3 The adopted values for the load distribution simulations of the BTS and the
+ 0.14 × 10 −6
× θfc − 273.15
(
(4) boost converter.
Load distribution parameters in the BTS
And the output voltage for the cells is obtained based on the
Battery voltage 48 V
following formulation: Number of BTS at the site 1
Total BTS load (Transceiver + microwave) 2380 W, 50 A
Vcell = Ea − ηma − ηal − ηO (5) Heat exchanger load 360 W
Microwave equipment load 100 W
This means that the output voltage is received after subtracting Load/Transceiver 12
the energy available from the internal losses inside the fuel cell, Transceiver 160 W
including mass transfer loss, ohmic loss, and activation loss. Parameters of boost converter
where, the activation loss is achieved as follows: Inductance 0.01 mH
[ ( )
pO2 Input voltage 26–39 V
ηal = − Ψ1 + Ψ2 θfc + Ψ3 θfc ln Output voltage 48 V
5.1 × 106 × exp (−498/θ ) Resistance 1.2 
Conductance 21.6 mF
]
+Ψ4 θfc ln (i) (6) fs 10 kHz

Since the presence of the electrical resistance in electrodes and

polymer membrane, Ohmic losses are also made by the following In this study, the state-space model of the system has been
equation: utilized so that two states model is employed due to two active
⎡ ⎛ ( (
θfc
) ( )
2.5 ( )) ⎞l ⎤ elements, i.e. the voltage of the capacitor and the current of the
181.6 1 + 0.062 × 303 × Ai + 0.03 × Ai
⎢ ⎟⎥ inductor.
ηO = ⎢Rc + ⎜
⎜
× i (7)
⎢ ⎥
]) ⎟
θ
( [ ⎠⎥
− 303
α β ρ
( ( ) ) [ ] [ ]
fc
⎝
i
A × λ − 3 × A − 0.63 + A × exp 4.2 ×
⎣ ⎦
θ fc ẋ = x+ V (9)
γ δ σ in
Finally, the mass transfer loss is obtained by the following equa-
y= τ ω x
[ ]
tion:
(
iden
) By considering the switch into state ‘‘on’’,
ηma = −β × ln 1 − (8) rL 1 1
ilimit ,den α = − , β = γ = σ = τ = 0, δ = − ,ρ = ,
L C (R + rc ) L
In the above cases, Ψ1 , Ψ2 , Ψ3 , Ψ4 , Rc are obtained by the opti- (10)
mization methods as described in Eslami et al. (2019). R
ω=
R + rc
4. Mathematical modeling and optimizing the boost converter And, by considering the switch into state ‘‘off’’,
RrC
Boost switching converter is a converter that increases the rL + R+rC R
α= , β = −γ = − ,
input voltage level to the output while reducing the current level. L C ( R + rc )
The booster converter is a class of switching power supplies
1 1 RrC
and typically comprises two semiconductors (one diode and one δ=− ,ρ = , σ = 0, τ = ,ω (11)
transistor) and at least one stator element (inductor or capacitor C (R + rc ) L R + rC
or a combination of both). To reduce the ripple voltage, capacitor
R
filters (sometimes combined with inductors) are added to the =
outputs and the inputs of these switching converters. R + rc
The boost power converter can be connected to any DC source A comprehensive model of the system is obtained by the averaged
such as fuel cell, battery, solar panel, rectifier, and DC generator. large-signal model within the circuit averaging when it is in ‘‘ON’’
The process of converting a DC voltage to another DC voltage state and ‘‘OFF’’ state of the switch over a cycle as:
at a different level is called DC/DC conversion. As the converter RrC
switched on, the source of the voltage and the inductor form a rL rL + R+rC
α= D+ (1 − D) ,
closed circuit. At this moment, the energy will be stored in the L L
indicator. After the circuit is switched off, this energy will be R
evacuated. This is why the output voltage converter is stable and β = −γ = − (1 − D) ,
C (R + rc )
has greater value than the input voltage. Then, in the feedback, (12)
the voltage level is sensed and sent to the comparator to find 1 1
δ=− ,ρ = ,σ
the error value. The error signal is then sent to a PI controller for C (R + rc ) L
correcting the output voltage. The reason for using PI controller
RrC R
instead of PID is that if the derivative term (D) is added to the = 0, τ = (1 − D) , ω =
boost control system, the presence of high-frequency noise in the R + rC R + rc
boost converter has been highly amplified. The derivative term where, DTS and (1 − D) TS represent ‘‘ON’’ and ‘‘OFF’’ states of the
can also be trouble maker during a sudden variation in the input switch.
reference such that If a big sudden variation happens, the men- Table 2 illustrates the adopted values for the load distribution
tioned momentary variation in the error e(t) that is multiplied by simulations of the BTS and the boost converter.
the derivative constant (KD ) will saturate the converter. The combination of the boost converter and the optimiza-
The modulator is finally adopted to form the optimal PWM tion algorithm forms the regulating unit. This unit adjusts the
signal for igniting the MOSFET. In this study, the optimal value source and the load voltages by increasing the voltage level.
of the PI controller is obtained by the proposed optimization Furthermore, the distribution DC bus for BTS requires a reliable,
algorithm. adjusted, and uninterrupted 48 V DC power source. This reliable
 D. Li, L. Deng, Q. Su et al. / Energy Reports 6 (2020) 297–307 301

voltage can be provided by the boost converter by considering distance based on time and intensity differences of sound wave
a low oscillation at the output. A significant case that should be arrival.
considered in the system is to design a reliable system that can
recovery itself after sudden disturbances with minimum time, 5.1. Owl search algorithm (OSA)
overshoot, and undershoot. This forces us to use a closed-loop
control system for this purpose.
Like any bio-inspired optimization algorithms, OSA starts with
a random set of the population as the initial solution of the
5. Barn owl and inspiration for optimization algorithm. The population in OSA shows the position of owls in a
forest as the search space. By considering n number of owls and d
In recent years, the application of bio-inspired techniques dimensional search space, the random position for owls is stored
is turned into popular global optimization methods. the main in n × d matrix as follows:
idea behind these techniques is to mimic different natural phe- ⎡ ⎤
nomena (Saeedi et al., 2019; Nejad et al., 2019; Mousavi and O1,1 ··· O1 ,d
Soleymani, 2014), social behavior (Gheydi et al., 2016; Aliniya and .
O = ⎣ .. .. .. ⎥
. (13)
⎢
. ⎦
Mirroshandel, 2019; Atashpaz-Gargari and Lucas, 2007; Razmjooy
On,1 ··· On,d
et al., 2011, 2013, 2017b), human competitions (Razmjooy et al.,
2016; Bandaghiri et al., 2016; Razmjooy et al., 2017a, 2018), etc. where, matrix element Oi,j describes the jth variable (dimension)
Intensification and diversification are important characteris- of ith owl. To generate a uniform distributed initial location, the
tics of the meta-heuristic methods. Intensification searches following formula has been applied:
around the current best solutions and selects the best candidate
points. Oi = Ol + (Ol + Ou ) × U (0, 1) (14)
The convergence ability and escaping from the local optimum where, U (0, 1) represents a uniformly distributed random integer
point are important parts of these techniques. Several new bio- in the interval 0 and 1, and Ou and Ol upper and lower bounds of
inspired algorithms are proposed for global search. Recently, Jain ith owl Oi in jth dimension, respectively.
et al. (2018) developed a promising bio-inspired technique, owl The cost for the owls’ location in a forest is calculated based
search algorithm (OSA). The OSA is based on the idealized behav- on a cost function and stored in a matrix as follows:
ior of the hunting characteristics of owls.
O1,1 , O1,2 , . . . , O1,d
⎡ ([ ]) ⎤
The owl is a bird that lives all over the globe except Antarctica. f1
These birds hunt at night, which is why they are known as night f=⎣
⎢ .. ⎥
(15)
. ⎦
predators. The owl is widely believed to be an infelicitous bird,
On,1 , On,2 , . . . , On,d
([ ])
fn
but it is a useful hunter because it can be used to kill rodents. The
eyes of owl are like a human being in front of his face. However, The cost value of the owls’ location here directly depends on
owls cannot move their eyes and have to turn their heads and the intensity information received through ears. In this condition,
neck to see their surroundings. maximum intensity (for maximization problems) can be obtained
The owl is an elongated owl, with wide wings, bright color, by the best owl as it is more close to prey. The normalized
and almost square tail. Depending on the subspecies, its length is intensity value information of ith owl is adopted for updating the
2 to 4 cm and its wing width is 1 to 2 cm. On the fly, they can position and may be achieved by:
be cleaned of ordinary owls by the movements of the waves and
fi − w
the legs covered with feathers. Ii = (16)
Its back surface is golden pea-colored and has tiny spots and b−w
has a fairly uniform white ventral surface. Its eyes are black and where,
he has no guppers. It usually hunts at night by sitting upright with
their long legs and big head. The owl’s face and legs are full of b = max fm (17)
m∈1,...,n
feathers. The male and female owls are similar, but the female
owl has a larger size than the male. The species vary in size and and,
can range from 15 cm to 76 cm. w = min fm (18)
The owl has very good hearing and vision. The owl’s visual m∈1,...,n
power is so high that it can help it hunting in the dark. Some The distance information for each owl and prey is formulated as
owls also hunt with their high hearing power. follows:
The barn owl is the most diverse species among different kinds √∑
of owls. Barn owls live in everywhere except poles and desserts Di = (Oi − L)2 (19)
from the alpine belt, large parts of Indonesia and the islands of
i
the ocean. It is a beautiful hunter bird at night, with a medium-
sized white and large feather, it has a short, square tail and long where L represents the prey location that is obtained by the fittest
legs and its legs are covered with white feathers. They have a owl.
great auditory system for locating the prey. Barn owls have grown The algorithm considers that there is only one prey (global op-
with a determined anatomical characteristic of auditory system timum) in the forest. In the hunting process, owls flights silent to-
by vertical asymmetry of ears. This specific characteristic is used wards the prey. The intensity changing for the ith owl is achieved
to obtain the sound in one ear before the other to exact locating below:
of prey (Grothe, 2018). Ii
The prey can be hidden in the dark by hearing sense instead Ci = + Rn (20)
Di 2
of vision sense (Moiseff et al., 2018). The generated sound by
prey is processed based on twp parts in the owl’s brain: the where, Di 2 is used instead of 4π Di 2 , and Rn is random noise to
interaural level (loudness) difference (ILD) and the interaural time make the model more realistic.
difference (ITD) which are used for preparing the auditory map Due to the prey movements, the owls need to change their
of prey location (Jain et al., 2018). The owl can estimate the prey present location silently. In OSA, the prey movement is modeled
302 D. Li, L. Deng, Q. Su et al. / Energy Reports 6 (2020) 297–307

based on probability and therefore new positions of owls are

achieved by the following position updating mechanism:

Oti + β × Ci × ⏐α L − Oti ⏐ , ppm < 0.5

{ ⏐ ⏐
Oti +1 = (21)
Oti − β × Ci × ⏐α L − Oti ⏐ , ppm ≥ 0.5
⏐ ⏐

where, ppm describes the probability of prey movement, α and β

present a uniformly distributed random number in the range [0,
0.5] and a linearly decreasing constant from 1.9 to 0, respectively.
Indeed, β is a large variation that develops the exploration
term of the search space. By developing these variations using
the algorithm, it can be reduced to improve the exploitation. The
OSA has only one parameter (β ) that makes it so reliable than the
other bio-inspired algorithms.

5.2. Improved balanced owl search algorithm (BOSA)

In recent years, by increasing the impact of nonlinear dy-

namics in system modeling, the application of chaos theory has
been increasing. Optimization is one of the fields which can be
impressed by the chaos theory.
From the standard OSA, the parameter β is the only random
value in the algorithm. Using random values in each iteration
makes a premature convergence for the system. In this study, Fig. 3. Block diagram of the proposed BOSA.
to cover this problem, a chaos mechanism, called singer map-
ping has been employed (Yang et al., 2007; Rim et al., 2018).
This mechanism maps the unknown parameter β into a regular
The first benchmark is Rastrigin with a dimension between 30
formulation:
and 50 and constraints in the range [−512, 512]:
βi+1 = 1.07 7.9βi − 23.3βi2 + 28.7βi3 − 13.3βi4
( )
(22) D
∑
f1 (x) = 10D + x2i − 10cos (2π xi )
( )
(27)
Another technique for improving the premature convergence in
i=1
OSA is to use Lévy flight (LF). LV is a popular mechanism that is
widely used in bio-inspired optimization algorithms to develop The second benchmark is Rosenbrock with a dimension between
premature convergence (Choi and Lee, 1998). The random walk is 30 and 50 and constraints in the range [−2.045, 2.045]:
the main part of this mechanism for proper handling of the local D−1
∑
f2 (x) = 100 x2i − xi+1 + (xi − 1)2
( ( ) )
search. This mechanism is mathematically formulated as follows: (28)
i=1
Le (w) ≈ w −1−τ
(23)
The third benchmark is Ackley. This benchmark has a dimension
A
w= (24) between 30 and 50 with the constraints in the range [−10, 10]
|B|1/τ which can be formulated as follows:
}2
Γ (1 + τ ) sin(πτ /2) τ
{  ⎛ ⎞
D

σ =
2
(25) 1 ∑
τ Γ ((1 + τ )/2) 2(1+τ )/2 f3 (x) = −20exp ⎝−0.2√
( )
x2i ⎠
D
i=1
where, τ describes the LV index in the range [0,2] (here, τ =
3/2 Li et al., 2018), A ∼ N(0, σ 2 ) and B ∼ N(0, σ 2 ), w is the step
( D
)
1∑
size, Γ (.) Determines Gamma function, A/B ∼ N(0, σ 2) means − exp (cos (2π xi )) + 20 + e (29)
D
that the samples produce from a Gaussian distribution in which i=1
mean is zero and variance is σ 2 , respectively. And the last benchmark is Sphere. This benchmark has a dimen-
Based on the explained cases, new positions of the owls can sion between 30 and 50 with the constraints in the range [−512,
be achieved by the following formulation: 512] which can be formulated as follows:
Oti + β × Ci × ⏐α L − Oti ⏐ × Le(δ ), ppm < 0.5
{ ⏐ ⏐ D
Oti +1
∑
= (26) f4 (x) = x2i (30)
Oti − β × Ci × ⏐α L − Oti ⏐ × Le(δ ), ppm ≥ 0.5
⏐ ⏐
i=1
The flowchart of the presented IOSA is shown in Fig. 3. Table 3 declares the simulation results of the presented method
compared with others.
The results show that using the proposed method gives the
5.3. The validation of the algorithm
best achievements compared with the other state of the art
methods.
To validate the proposed algorithm, four standard benchmarks
have been analyzed. The algorithm is compared with some new 6. The main architecture of the optimized system for the tele-
algorithms including emperor penguin optimization (EPO) (Dhi- com supply system
man and Kumar, 2018), shark smell optimization (SSO) algo-
rithm (Abedinia et al., 2018), world cup optimization algorithm The main structure of the supply system includes a boost
(WCO) (Razmjooy et al., 2016), and the original OSA. converter, a modulator, a comparator, and a controller. MOSFET
 D. Li, L. Deng, Q. Su et al. / Energy Reports 6 (2020) 297–307 303

Table 3
The obtained results of the different algorithms for validation.
Benchmark IOSA EPO GA (Nejad et al., 2019) SSO (Bansal, 2019) WCO (Razmjooy et al., 2016) OSA
f1 MD 0.00 2.28 70.61 74.24 2.19 3.15
SD 0.00 3.53 1.66 8.96 4.35 3.74
f2 MD 4.28 13.10 35.41 200.1 13.16 12.95
SD 3.62 4.72 27.15 59.00 4.62 7.6
f3 MD 0.00 6.32e−16 3.19e−2 8.26 3.14e−3 4.85e−16
SD 0.00 0.00 2.14e−2 1.19 1.12e−3 0.00
f4 MD 0.00 0.00 1.15e−4 8.27e−4 6.19e−9 0.00
SD 0.00 0.00 3.14e−5 5.12e−4 3.28e−9 0.00

Table 4 Fig. 4 shows the architecture of the boost converter by feedback

Parameter setting of the algorithms. method including a comparator, modulator, and PI controller.
Algorithm Parameter Value Ziegler–Nichols (ZN) is a method that steps up the propor-
No. of particles 100 tional coefficient so long as the feedback system becomes criti-
No. of generation 1000
PSO cally stable. However using ZN as a classic method for adjusting
c1 and c2 2
Inertia weight 0.9
the controller coefficient is a good selection; it sometimes stuck in
the local minimum. To solve this problem, as aforementioned in
Search agents 100
No. of generation 1000 this study, a new optimization technique, called IOSA is proposed
Temperature profile 500 for achieving a global minimum value of the error signal for
⃗ Constant
A [−1.5, 1.5] the boost converter. The method has also the ability to pre-
EPO
Function S () [0,1.5] vent premature convergence. The cost function for the converter
Parameter M 2
optimization in this research is declared below:
Parameter f 2.5 ∫ ∞
Parameter l 1.5 1
min F (KP , KI ) = t × E(t)2 dt (33)
Search agents 100 t
IOSA 0
No. of generation 1000
Subject to,
KP ,min , KI ,min ≤ KP , KI ≤ KP ,max , KI ,max (34)
is used here for tuning the boost converter based on switching on And the cost of the fitness function is:
and switching off. 1
As aforementioned, to proper control of the PEMFC in the Cost function = (35)
1 + F (KP , KI )
base transceiver station, the feedback signal of the output voltage
(VO (t)) is sensed and compared with the desired voltage (VD (t)) where, θ = f (KP , KI ).
to detect the value of the error signal, i.e.
7. System analysis
E (t ) = VD (t) − VO (t) (31)
As before mentioned, this study analyzes a new supply con-
Afterward, the op-amp receives the error signal to operate as PI figuration in the presence of PEMFC for base transceiver station
controller. To achieve optimal coefficients (the proportional (KP ) in a remote area. The proposed architecture has two main ad-
and the integral (KI ) coefficients) for the controller, IOSA has been vantages. Firstly, using this system reduces the requirement for
utilized, such that: multiple batteries package. Secondly, this configuration provides
∫ green energy instead of diesel generator. The study utilized a new
optimized configuration of the boost converter based on a new
u (t ) = Kp + KI E(t) (32)
algorithm, called BOSA, to generate an interface between PEMFC
and BTS load to achieve a regulator with quick dynamic response.
The reason for selection the OSA was its results for global op- In this study, a 30 V, 1.8 kW PEMFC is considered in a standard
timization in comparison with the other new introduced algo- condition. Nevertheless, changes in the load and environment
rithms. This is also proved after applying its improved version to such as temperature and humidity put the output voltage of the
the Telecom system. It is important to note that this algorithm PEMFC in the range 26 V and 39 V. In addition, changes in the
gives good results for our case study, in other words, it is not number of the transceivers make some variations on the power
guaranteed to give the best results for other applications which of the telecom load in the range 500 W and 2200 W. Therefore,
shows the nature of the meta-heuristics. To present an impartial the main purpose is to provide a regulator to achieve a stable
evaluation, common parameters of all the algorithms including voltage for the distribution bus of the BTS in 48 V by considering
the number of iteration (1000) and the population are considered the aforesaid variations.
to be the same. More setting are given in Table 4. In this study, the efficiency of the presented optimized system
The standard parametric settings of algorithms are used as based on IOSA for regulating the BTS supply is compared with
given in Table 2. A valid statistical analysis is performed by the classical PI controller, i.e. ZN method and also the standard
independently executing each algorithm for 30 trial runs for each OSA and Emperor Penguin Optimizer (EPO) (Dhiman and Ku-
benchmark function with maximum 30 000 number of function mar, 2018) as one of the newly proposed optimization algorithm
evaluations in each run. The recorded results are presented in to have a fairy analysis. Three different scenarios have been
Table 3 in terms of their best, worst, mean value and standard validated for the system analysis as follows:
deviation (SD).
Then, the modulator applies an optimal PWM signal control for (1) Analysis of the tracking ability of the output voltage from
igniting the MOSFET. This process makes an optimal configuration the desired voltage (Vd ) with ±10% variations for the volt-
for the boost converter in the DC distribution bus of the BTS. age of the distribution bus, i.e. [43 V, 53 V].
304 D. Li, L. Deng, Q. Su et al. / Energy Reports 6 (2020) 297–307

Fig. 4. The architecture of the boost converter by feedback method and PI controller.

Table 5
The regulator PI parameters achieved by different methods.
Fitness function Algorithm KP KI
ZN 0.0406 8.12
BOSA 6.3 62.81
MSE
EPO 5.7 71.4
PSO 1.6 88.3

Table 6
A detailed study on the system dynamic response.
Dynamic response characteristic ZN IOSA EPO PSO
Settling time (ms) 96.37 10 35 84
Rise time (ms) 1.2 1.1 1.2 0.3
Overshoot (%) – – 20 –
Peak time (ms) 4 5 18 2

method has the worst results such that it has about 20% under-
shoot and it decreases to about 29.5 V at that time. This shows
Fig. 5. The dynamic response of the proposed IOSA compared with the analyzed
methods during an increasing input voltage from 0 to 30 V.
how the optimization algorithms can improve system efficiency.
It can be also observed that the results of the proposed IOSA
regulator have no undershoot or overshoots in this condition.
Settling time for this method is about 5 ms. Table 5 illustrates
(2) Analysis of the disturbance rejection ability of the regulator
the PI parameters that are achieved by the proposed method and
by considering the input voltage (Vin ) changes in the boost
the ZN, PSO, and EPO methods and Table 6 declares the system
converter in the range [26 V, 39 V].
dynamic response based on these values.
(3) Analysis of the disturbance rejection in the output by con-
In Fig. 6, we consider an input voltage step up from 26 V to
sidering the controller efficiency for 22% of overloading and
39 V at time 0.5 s. The settling time of the introduced IOSA-based
72% of under loading condition when the peak traffic of the
regulator is about 48.82 ms by an overshoot of less than 5%. Here,
communication signals and when only a transceiver works,
using ZN based regulator, PSO based regulator, and EPO based
respectively.
regulator gives 51.4 s, 49.4 s, and 49.9 s, respectively.
In the following, a reverse state has been considered. By as-
suming a drop down for the input voltage from 39 V to 26 V
at time = 0.5 s, the settling time for the introduced BOSA, ZN,
Simulation results have been applied by the Matlab R2017b plat- PSO, and EPO based regulators are 42.54 ms, 60.5 ms, 53 ms,
form. In this part, the proposed method has been compared with and 44.6 ms, respectively. In addition, the output of the proposed
ZN method as a classic controller and also PSO algorithm and EPO IOSA is the closest results to the desired voltage during the input
algorithm (Dhiman and Kumar, 2018) as optimized algorithms. perturbation (see Fig. 7).
At first, consider a voltage step-up from 0 V to 30 V whereas In Fig. 8, the load oscillation for the studied system in about
the purpose is to track the desired voltage (Vd = 49 V) of 1.8 kW 4.6  has been shown.
load. Fig. 5 shows the dynamic response of the proposed IOSA The results of this variation are shown in Fig. 9. It can be seen
based system compared with the analyzed methods. that the dynamic response of the boost converter by ZN regulator
As can be seen, the proposed IOSA technique has the fastest
results for the studied case study while the ZN method as a classic
 D. Li, L. Deng, Q. Su et al. / Energy Reports 6 (2020) 297–307 305

Fig. 6. The dynamic response of the proposed IOSA compared with the analyzed Fig. 9. The output voltage of the system by the proposed IOSA compared with
methods during an increasing input voltage from 0 to 30 V. the analyzed methods for optimizing boost converter for the load variation at
time 0.5 s.

Fig. 10. The load variations for the studied system in about 1.2 .

Fig. 7. The dynamic response of the proposed IOSA compared with the analyzed
methods during an increasing input voltage from 30 to 0 V.

Fig. 8. The load variations for the studied system in about 4.6 . Fig. 11. The output voltage of the proposed IOSA compared with the analyzed
methods for load variation at time 0.5 s.

indicates an extra undershoot to the 28.7 V at the time 30.2 ms.

Other methods have good results for this case. respectively. The output of the proposed method gives the best
In Fig. 10, the load oscillation for the studied system in about results compare with the other methods.
1.2  has been shown and the results are shown in Fig. 11. Consider another condition with a value of ±10% oscillations
As can be seen, the settling time for the proposed IOSA, ZN, from the desired voltage which is applied at the time 0.5 s.
EPO, and PSO is achieved 56 ms, 56 ms, 56 ms, and 46.61 ms, Figs. 12 and 13 show 53 V and 48 V load variation at time 0.5
s, respectively. It is observed that among different methods, the
306 D. Li, L. Deng, Q. Su et al. / Energy Reports 6 (2020) 297–307

Declaration of competing interest

The authors declare that they have no known competing finan-

cial interests or personal relationships that could have appeared
to influence the work reported in this paper.

CRediT authorship contribution statement

Daming Li: Conceptualization, Data curation, Writing - origi-

nal draft, Writing - review & editing. Lianbing Deng: Conceptu-
alization, Data curation, Writing - original draft, Writing - review
& editing. Qinglang Su: Conceptualization, Data curation, Writing
- original draft, Writing - review & editing. Yonghua Song: Con-
ceptualization, Data curation, Writing - original draft, Writing -
review & editing.

Acknowledgments

Fig. 12. Step response for ±10% oscillations of the output voltage in the
1. Project funded by China Postdoctoral Science Foundation.
proposed IOSA compared with the analyzed methods.
2. Project funded by National Key R&D Program of China (No.
2017YFB0503604 ; No. 2017YFB0503801).
3. Project funded by Electronic fence system project, China.
4. Project funded by the Project (0172018A) of FDCT, China.
5. Project funded by the Project of Macao Foundation, China.

References

Abedinia, O., Amjady, N., Ghadimi, N., 2018. Solar energy forecasting based
on hybrid neural network and improved metaheuristic algorithm. Comput.
Intell. 34, 241–260.
Aghajani, Gholamreza, Ghadimi, Noradin, 2018. Multi-objective energy manage-
ment in a micro-grid. Energy Rep. 4, 218–225.
Akbary, Paria, et al., 2019. Extracting appropriate nodal marginal prices for all
types of committed reserve. Comput. Econ. 53 (1), 1–26.
Aliniya, Z., Mirroshandel, S.A., 2019. A novel combinatorial merge-split approach
for automatic clustering using imperialist competitive algorithm. Expert Syst.
Appl. 117, 243–266.
Atashpaz-Gargari, E., Lucas, C., 2007. Imperialist competitive algorithm: an algo-
rithm for optimization inspired by imperialistic competition. In: Evolutionary
Computation, 2007. CEC 2007. IEEE Congress on. pp. 4661–4667.
Bagal, Hamid Asadi, et al., 2018. Risk-assessment of photovoltaic-wind-battery-
Fig. 13. Step response for ±10% oscillations of the output voltage in the grid based large industrial consumer using information gap decision theory.
proposed IOSA compared with the analyzed methods. Sol. Energy 169, 343–352.
Bandaghiri, P.S., Moradi, N., Tehrani, S.S., 2016. Optimal tuning of PID controller
parameters for speed control of DC motor based on world cup optimization
algorithm. Parameters 1, 2.
ZN based regulator has the most delay for the dynamic response Bansal, J.C., 2019. Particle swarm optimization. In: Evolutionary and Swarm
while the proposed IOSA has the best results. Intelligence Algorithms. ed: Springer, pp. 11–23.
Choi, C., Lee, J.-J., 1998. Chaotic local search algorithm. Artif. Life Robot. 2, 41–47.
Choi, Y.B., Lee, J., Yoo, S., Yoo, Y.K., 2019. Value relevance of customer equity be-
8. Conclusion
yond financial statements: evidence from mobile telecom industry. Asia-Pac.
J. Account. Econ. 26, 281–300.
This study presents a new optimized method for PEMFC based Dhiman, G., Kumar, V., 2018. Emperor penguin optimizer: A bio-inspired
energy conversion system to supply guaranteed energy for a algorithm for engineering problems. Knowl.-Based Syst. 159, 20–50.
telecom system. The main objective is to develop the system effi- Ebrahimian, Homayoun, et al., 2018. The price prediction for the energy market
based on a new method. Econ. Res.-Ekon. Istraž. 31 (1), 313–337.
ciency to decrease the operating cost. By considering the PEMFC Eslami, Mahdiyeh, et al., 2019. A new formulation to reduce the number of
along with DC distribution bus with output voltage regulation, variables and constraints to expedite SCUC in bulky power systems. Proc.
the reliability and the power supply quality have been increased Natl. Acad. Sci. India A 89 (2), 311–321.
and make it interruptible to the loads. The regulator is designed Firouz, Mansour Hosseini, Ghadimi, Noradin, 2016. Concordant controllers based
on FACTS and FPSS for solving wide-area in multi-machine power system. J.
based on the PI technique to hold the output voltage of the boost
Intell. Fuzzy Systems 30 (2), 845–859.
converter in the 48 V. A new improved version of the Owl search Gheydi, Milad, Nouri, Alireza, Ghadimi, Noradin, 2016. Planning in microgrids
algorithm based on Singer mapping and Levy flight is proposed with conservation of voltage reduction. IEEE Syst. J. 12 (3), 2782–2790.
to improve the PI controller. For the performance analysis of the Gollou, Abbas Rahimi, Ghadimi, Noradin, 2017. A new feature selection and
hybrid forecast engine for day-ahead price forecasting of electricity markets.
proposed method, it has been compared with a classical method,
J. Intell. Fuzzy Systems 32 (6), 4031–4045.
Ziegler–Nichols based controller and some optimization based Grothe, B., 2018. How the barn owl computes auditory space. Trends Neurosci.
techniques including particle swarm optimization and emperor 41, 115–117.
penguin optimizer based PI controllers to show the system per- Hallur, G.G., Sane, V.S., 2018. Indian telecom regulatory framework in comparison
formance that shows the excellence of the proposed method as with five countries: structure, role description and funding, digital policy.
Regul. Gov. 20, 62–77.
an optimal regulator in the base transceiver station. In the future Hamian, Melika, et al., 2018. A framework to expedite joint energy-reserve
work, a combined cooling, heating, and power system along with payment cost minimization using a custom-designed method based on
Telecom tower will be proposed to improve the system efficiency. mixed integer genetic algorithm. Eng. Appl. Artif. Intell. 72, 203–212.
 D. Li, L. Deng, Q. Su et al. / Energy Reports 6 (2020) 297–307 307

Hosseini Firouz, Mansour, Ghadimi, Noradin, 2016. Optimal preventive mainte- Nejad, Hadi Chahkandi, et al., 2019. Reliability based optimal allocation of
nance policy for electric power distribution systems based on the fuzzy AHP distributed generations in transmission systems under demand response
methods. Complexity 21 (6), 70–88. program. Electr. Power Syst. Res. 176, 105952.
Jain, M., Maurya, S., Rani, A., Singh, V., 2018. Owl search algorithm: a novel Ojo, J.S., Owolawi, P.A., Atoye, A.M., 2019. Designing a green power delivery
nature-inspired heuristic paradigm for global optimization. J. Intell. Fuzzy system for base transceiver stations in southwestern Nigeria. SAIEE Afr. Res.
Systems 34, 1573–1582. J. 110, 19–25.
Kaur, R., Krishnasamy, V., Muthusamy, K., Chinnamuthan, P., 2017. A novel Paudel, P., Wasti, S., 2018. Peak demand management in micro hydro using
proton exchange membrane fuel cell based power conversion system battery bank. Hydro Nepal: J. Water Energy Environ. 22, 34–40.
for telecom supply with genetic algorithm assisted intelligent interfacing Razmjooy, N., Khalilpour, M., Ramezani, M., 2016. A new meta-heuristic opti-
converter. Energy Convers. Manag. 136, 173–183. mization algorithm inspired by FIFA world cup competitions: Theory and its
Khodaei, Hossein, et al., 2018. Fuzzy-based heat and power hub models application in PID designing for AVR system. J. Control Autom. Electr. Syst.
for cost-emission operation of an industrial consumer using compromise 27, 419–440.
programming. Appl. Therm. Eng. 137, 395–405. Razmjooy, N., Madadi, A., Ramezani, M., 2017a. Robust control of power system
Leng, Hua, et al., 2018. A new wind power prediction method based on ridgelet stabilizer using world cup optimization algorithm. Int. J. Inform. Secur. Syst.
transforms, hybrid feature selection and closed-loop forecasting. Adv. Eng. Manag. 5, 7.
Inform. 36, 20–30. Razmjooy, N., Mousavi, B.S., Sadeghi, B., Khalilpour, M., 2011. Image thresholding
Li, X., Niu, P., Liu, J., 2018. Combustion optimization of a boiler based on the optimization based on imperialist competitive algorithm. In: 3rd Iranian
chaos and Levy flight vortex search algorithm. Appl. Math. Model. 58, 3–18. Conference on Electrical and Electronics Engineering (ICEEE2011).
Liu, Yang, Wang, Wei, Ghadimi, Noradin, 2017. Electricity load forecasting by an Razmjooy, N., Mousavi, B.S., Soleymani, F., 2013. A hybrid neural network
improved forecast engine for building level consumers. Energy 139, 18–30. imperialist competitive algorithm for skin color segmentation. Math. Comput.
Mahajan, V., Mahajan, R., Misra, R., 2019. Indian telecom industry: Challenges Modelling 57, 848–856.
and use of analytics to manage customer churn. In: Handbook of Research on Razmjooy, N., Ramezani, M., Ghadimi, N., 2017b. Imperialist competi-
Expanding Business Opportunities with Information Systems and Analytics. tive algorithm-based optimization of neuro-fuzzy system parameters for
ed: IGI Global, pp. 353–376. automatic red-eye removal. Int. J. Fuzzy Syst. 19, 1144–1156.
Mirzapour, Farzaneh, et al., 2019. A new prediction model of battery and wind- Razmjooy, N., Sheykhahmad, F.R., Ghadimi, N., 2018. A hybrid neural network–
solar output in hybrid power system. J. Ambient Intell. Humaniz. Comput. world cup optimization algorithm for melanoma detection. Open Med. 13,
10 (1), 77–87. 9–16.
Moiseff, A., Haresign, T., Wang, J., 2018. Sound localization from binaural cues Rim, C., Piao, S., Li, G., Pak, U., 2018. A niching chaos optimization algorithm for
by the barn owl auditory system. In: Neuroethological Studies of Cognitive multimodal optimization. Soft Comput. 22, 621–633.
and Perceptual Processes. ed: Routledge, pp. 305–323. Saeedi, Mohammadhossein, et al., 2019. Robust optimization based optimal
Mousavi, B.S., Soleymani, F., 2014. Semantic image classification by genetic chiller loading under cooling demand uncertainty. Appl. Therm. Eng. 148,
algorithm using optimised fuzzy system based on Zernike moments. Signal 1081–1091.
Image Video Process. 8, 831–842. Yang, D., Li, G., Cheng, G., 2007. On the efficiency of chaos optimization
algorithms for global optimization. Chaos Solitons Fractals 34, 1366–1375.