sensors

Article
Power Factor Compensation Using Teaching Learning
Based Optimization and Monitoring System by Cloud
Data Logger
Antonio Cano Ortega * , Francisco Jose Sánchez Sutil and Jesús De la Casa Hernández
Department of Electrical Engineering, University of Jaen, 23071 EPS Jaen, Spain; fssutil@ujaen.es (F.J.S.S.);
jcasa@ujaen.es (J.D.l.C.H.)
* Correspondence: acano@ujaen.es; Tel.: +34-953-212343




Received: 2 April 2019; Accepted: 7 May 2019; Published: 10 May 2019


Abstract: The main objective of this paper is to compensate power factor using teaching learning
based optimization (TLBO), determine the capacitor bank optimization (CBO) algorithm, and monitor
a system in real-time using cloud data logging (CDL). Implemented Power Factor Compensation and
Monitoring System (PFCMS) calculates the optimal capacitor combination to improve power factor of
the installation by measure of voltage, current, and active power. CBO algorithm determines the best
solution of capacitor values to install, by applying TLBO in different phases of the algorithm. Electrical
variables acquired by the sensors and the variables calculated are stored in CDL using Google Sheets
(GS) to monitor and analyse the installation by means of a TLBO algorithm implemented in PFCMS,
that optimizes the compensation power factor of installation and determining which capacitors are
connected in real time. Moreover, the optimization of the power factor in facilities means economic
and energy savings, as well as the improvement of the quality of the operation of the installation.

Keywords: Power factor compensation monitoring system PFCMS; power factor PF; Cloud data
logger CDL; Teaching learning based optimization TLBO; capacitor bank optimization CBO

1. Introduction
Differently from residential loads, most commercial and industrial premises have a high uptake
of inductive loads such as electric motors, inductive/resistor load, sodium vapour, and metal halide
lighting, etc. These installations and operation of these devices distort power supply and reduce power
factor. Normally, a lower power factor for facilities causes a huge amount of losses and may lead to
a thermal problem in switchgear. Fortunately, power factor is controllable with properly designed
power factor improvement methods.
At present, society is involved in a major digital transformation due to major advances in
information and communication technologies. This digital transformation extends to the technological
sector, offering very interesting development possibilities. New technologies developed include the
Internet of things, big data, cloud computing, industry 4.0, and intelligent networks.
The application of new technologies to industry 4.0 allows monitoring the operating of the system
and controlling the system with computers and mobile devices remotely. In this sense, the measurement
and monitoring of electrical variables is interesting for studying the energy behaviour of installations.
Excessive reactive energy consumption implies an increase in the electricity bill, so it must be reduced
to achieve significant energy and economic savings.
To achieve energy and economic savings by correcting the power factor of the installation, it
is necessary to solve an optimization problem. There are many optimization algorithms applied
to the resolution of engineering problems, such as the Genetic Algorithm (GA), Particle Swarm

Sensors 2019, 19, 2172; doi:10.3390/s19092172 www.mdpi.com/journal/sensors

Sensors 2019, 19, 2172 2 of 23

Optimization (PSO), Artificial Bee Colony (ABC), etc. The most commonly used evolutionary
optimization technique is the genetic algorithm (GA). However, GA provides a near optimal solution
for a complex problem having a large number of variables and constraints. Parameters such as
population size, crossover ratio, mutation ratio are difficult to determine, this being the main difficulty
in applying this algorithm, which influences the effectiveness of the algorithm. Similarly, PSO and
ABC require others configuration parameters.
The main reason to use an optimization algorithm is its capacity to solve different optimization
problems effectively and efficiently. In this paper, TLBO is proposed to obtain global solutions for
continuous non-linear functions with less computational effort and high consistency. The TLBO
method is based on the effect of the influence of a teacher on the output of learners in a class.
Rao et al. [1] compared different optimization methods, which shows the advantage of using
the TLBO algorithm over other optimization algorithms (GA, PSO, ABC, etc.); different experiments
have been carried out to test the effectiveness of the TLBO against other optimization techniques, with
different objective functions.
The TLBO algorithm applies in [2], developing a new advanced TLBO algorithm to process
parameter optimization in selected modern machining. In electrical systems, TLBO is used to solve
different problems, such as obtain the optimal power flow (OPF) by means of a non-domination based
sorting multiobjective presented in [3]. In [4–6], the TLBO approach is used to minimize power loss
and energy cost by optimal placement of capacitors in radial distribution systems.
Moreover, there are different technologies for power factor correction (PFC), an overview of
the state of the art in reactive power compensation technologies, the principles of operation, design
characteristics of reactive power compensators implemented with thyristors, and self-commutated
converters are presented in [7]. Another paper [8], realizes an analysis and compensator design
framework for power-factor compensation based on cyclodissipativity. The challenges and power
quality issues faced in the micro grid are proposed in [9] by a review of compensation methods that
achieve an improved power factor and real power balance from the load point of view. The technical
and economic value of power factor improvement, verified through an analysis of a real-world electrical
system and loads, is implemented in [10].
Numerous investigations have been done on PFC in electrical installations. In this sense, in [11] a
new digital PFC control strategy overcomes the problem of limited switching frequency due to a limited
digital signal processor speed. In [12] a Fuzzy logic controlled synchronous motor is used for reactive
power compensation. A fuzzy logic controlled synchronous motor can give a very fast response to the
reactive power required by the load. Another approach to compensate for reactive energy is studied
in [13], that uses a Programmable Logic Controller (PLC) based PFC method for a 3-phase Induction
Motor (IM) through switching of shunt capacitors. In this context [14], the correction method use an
intelligent PFC based on PLC as a control system especially for correction of the industrial power factor.
Other PFC methods develop different compensation systems, for example in [15] a static var
compensator prototype was used to test and validate a variety of control strategies. An automatic
PFC unit was developed in [16], which is able to monitor the energy consumption of a system and
automatically improve its power factor, the device calculates the reactive power consumed by a
system’s inductive load and compensates the lagging power factor using capacitance from a capacitor
bank (CB).
In AC/DC conversion systems it is necessary to compensate the power factor. The paper [17] analyses
a control algorithm of a three-phase three-level PFC rectifier, using a mathematical analysis of two-level
space vector modulation. In [18], a system for reactive power compensation for battery/photovoltaic
hybrid power source of hybrid electric vehicle in real time is presented. In this context, a predictive
algorithm for AC–DC three-phase converters with active PFC is presented in [19].
The normative section includes [20–22], the IEEE Standards 18-2012, C37.26-2014, and C37.99-2012,
in which are developed the IEEE standards for shunt power capacitors, the protection of shunt CB´s,
and the methods of power-factor measurement for low-voltage inductive test circuits.
Sensors 2019, 19, 2172 3 of 23

A multitude of papers have emerged utilizing smart meter data, as a new smart voltage and
current monitoring system technique proposed in [23,24], as well as a smart home electric energy
saving system implemented by combining smart meter, smart plug, smart mobile devices, and database
server. The meters measure consumption on a very fine scale, usually on a 15 min basis, and the data
giving unprecedented granularity of consumption patterns at household level is analysed in [25].
This paper presents the development and implementation of PFCMS, using the open source
Arduino platform where the TLBO algorithm is implemented to operate in real time. Therefore, the
interest of the proposal is based on the advantage of the Arduino platform with a highly efficient
algorithm such as TLBO, in order to reduce the calculation and operation times of PFCMS.
In the first phase, the design of a CB involves determining the number and capacity of the
capacitors. For this, there should be a set of measurements of the electrical variables of the installation
and a comprehensive analysis of the installed loads. In this research, the algorithm CBO developed
calculates the optimal capacity of all capacitors to install. The number of capacitors influences the
complexity of the problem decisively, in this sense to evaluate the size of the capacitors, TLBO algorithm
reduces simulation times. CBO algorithm uses TLBO as the basis for optimization. The CBO algorithm
has to be programmed in a high-level language, such as C, Python, MatLab, etc. This research uses
MatLab as a programming environment.
The second phase, starting from the results of the optimal capacity to be installed, the compensation
of the power factor is done using TLBO in real time. As proof, there was analysis with PFCMS: a series
of tests were performed with different types of loads, such as lamps, motors and resistive-inductive
loads, where the electrical variables of the installation are measured and monitored, using the sensors
installed in PFCMS. Once the electrical variables were obtained, the TLBO algorithm was applied that
determined the combination of capacitors at each moment in order to obtain the power factor nearest
to the desired value. The experimental results demonstrate the feasibility of the proposal.
The measurement time can be adjusted from 0.2 to the desired time. Moreover, reduced
measurement time, together with the speed of resolution of TLBO, allows using PFCMS in real time.
Therefore, TLBO complies with the definition of the problem to be solved in this investigation.
The design of a CB involves determining the number and capacity of the capacitors. For this, there
should be a set of measurements of the electrical variables of the installation and a comprehensive
analysis of the installed loads. In this research the algorithm CBO developed calculates the optimal
capacity of all capacitors to install. The number of capacitors influences the complexity of the problem
decisively, in this sense to evaluate the size of the capacitors, TLBO algorithm reduces simulation times.
CBO algorithm uses TLBO as the basis for optimization.
It is possible to integrate the PFCMS with Android applications to be used with mobile devices
under this operating system. It is specially designed to work with an app developed for this type of
device. The apps for Android devices have been developed with the free programming environment
of Google MITAI [26].
All software development has been done with free programming environments, so that the system
can be reproduced by any researcher without the need to acquire commercial software licenses.
To make the measurements of the electrical variables, the electric power meter PZEM-004t
(PZEM) [27] (Ningbo Peacefair Electronic) was used. The measured variables are voltage, current, active
power, and energy, from which reactive, apparent power and PF are derived, for system monitoring.
The aim of this work is to design and implement a power factor compensation system using a
TLBO algorithm in real time, additionally developing an algorithm to obtain the optimal capacity of
the CB, monitor the system in GS, and a smartphone application to monitor the installation in real
time. PFCMS has been designed to correct power factor in different electric systems. The paper is
organized as follows: Section 2 presents the relevant related theory power factor, algorithm TLBO,
and optimal capacity algorithm, Section 3 deals with the hardware and software development of
PFCMS. Section 4 defines the case studies and results. Finally Section 5 exposes the conclusion of the
present investigation.
Sensors 2019, 19, 2172 4 of 23

2. Theoretical Background

2.1. Introduction to Compensation Power Factor

Power factor is the ratio between the active power P and the apparent power S in an electrical
load. It is simply a measure of how efficiently the load current is being converted into useful work
output. The lower the power factor of a system, the less economically it operates. A low power factor
can be the result of a significant phase difference between voltage and current at load terminals.
Generally, it is the use of inductive loads such as IM, power transformers, induction furnaces, and
so on that causes a current to lag behind voltage. A poor power factor resulting from inductive loads
can be improved by power factor correction method. Since power factor in inductive loads is generally
lower, they have to be supplied with reactive power in order to reduce increased power consumption
of the facilities.
All inductive loads require P to perform the actual work, and reactive power Q to maintain the
magnetic field. This Q is necessary for the equipment to operate, but imposes an undesirable weight
on the supply, causing the current to be out of phase with the voltage (current lags the voltage). Low
power factor can also result when inactive motors operate at less than full load, etc.
PFC is applied to neutralize as much of the magnetizing current as possible and to reduce losses
in the distribution systems [28–30]. A new approach for real time voltage control of distribution
networks that has improvements over the conventional voltage control models is [31]. It offers many
benefits to the commercial electrical consumer, including reduced utility bills by eliminating charges
on reactive power, reduced losses making extra S available from the existing supply. Thus, it improves
energy efficiency.

2.2. Compensation Power Factor Theory

The design of the device for compensating the adverse effects of inductive Q is a crucial task in
designing any PFC systems. The capacitive current supplied by the capacitors is directly connected
across the industrial load terminals or electrical installation. The consumer advises to improve the
power factor beyond the value of magnetizing kVAr rating of the load [14].
Lower power factors can dramatically increase the required current being consumed by an
appliance to work correctly. The following equation calculates the amount of reactive power that is
wanted to be produced by the bank of capacitors connected in parallel to the load:

QC = P × tan ϕV − ϕI ) − tan(ϕV − ϕI )0
h  i
(1)

where ϕV is the angle of the voltage signal and ϕI is the angle of the current signal, and QC is the
reactive power the capacitor must provide; and (ϕV − ϕI )0 is the angle we want for the power factor
once it is corrected.
This involves a detailed survey of loads connected to the system and performing measurements
of the electrical variables of the facilities during that period that allows us to determine the optimum
capacity of the CB. The QC requirement value changes with the load variation, hence the algorithm
control developed such that the reactive power requirement near the load terminals is maintained at a
constant by varying the capacitive reactive power and the power factor is maintained nearer to unity.
Knowing that the resistance of a capacitance is dependable on the frequency, it is known that the
capacitance needed to correct a certain amount of reactive power in single phase is

QC
C= (2)
2 × π × f × V2

where C is the capacitance in Farads, f is the frequency, and V is the voltage.

Sensors 2019, 19, 2172 5 of 23

In the three-phase systems, it can be connected to CB in wye (W) and delta (D). In the case of the
wye connection, the voltage phase to phase VPP is the voltage per phase by the square root of 3, and
the equation will be
QC
Cλ = (3)
2 × π × f × VPP
2

In the three-phase system connected in delta, the voltage phase to phase is equal to the voltage
per phase, the equation will be
QC
C∆ = (4)
3 × 2 × π × f × VPP
2

2.3. TLBO Algorithm

TLBO is a global stochastic optimization algorithm, based on populations and oriented to
large-scale problems. It is based on modeling the behavior of a class of students formed by a set of
known candidate solutions [32].
These solutions are progressively improved by simulating both the teaching process of a teacher
and the interaction between students. The main characteristic of TLBO is its performance and its lack
of specific search parameters, since it is only necessary to specify the population size and the number
of cycles as required.
To achieve this, TLBO relies on two fundamental steps per cycle, the phase of the teacher (Teacher
Phase (TP)) and that of the students (Learners Phase (LP)).
The TP difference between the respective mean result of each subject and the corresponding result
of the teacher for each subject is

Xnew = Xi + r×(Xteacher − TF × Xmean ) (5)

where, Xteacher is the best individual, Xi is other individual, Xmean is the current mean of the individuals,
TF is the teaching factor which decides the value of mean, and Xnew is the influence for student Xi by
the difference between the teacher´s knowledge and the qualities of all students.
In the LP, two possible solutions are selected, each final individual i from the previous stage is
paired with another, different from himself, j. Next, we try to move the individual i from his current
position in a direction that depends on its relative value with respect to j, finally, the individual i will
only be updated if his value is improved after the change.
 
Xnew = Xi + r × Xi − X j (6)
 
Xnew = Xi + r × X j − Xi (7)

Problem Formulation
The objective of optimal PFC problem of electrical installations is to optimize the objective function,
defined as the difference between power factor desired (PFDES ) and power factor obtained (PFOBT )
while satisfying all operational constraints.
The optimization function used to obtain the compensation of the power factor in real time is
given by  
2 
P P 
n 1
 −1  q(t) − j i=0 C j ∗ i × 2 × π × f × v (t) 
 

F = cos(ϕdes ) − costan   (8)
   p(t) 

where cos(ϕdes ) is the PFDES , q(t) is the reactive power, p(t) is the active power, and v(t) is the voltage of
the installation for a time t, Cj are the capacities of the capacitors and f is the frequency of the installation.
Sensors 2019, 19, 2172 6 of 23

The constraint used in the formulation of the problem for the developed TLBO algorithm has
been taken:
Sensors 2019, 19, x FOR PEER REVIEW 0.95 ≤ cos ϕ ≤ 1 6 of(9)
24

PFC depend on the current legislation in each country. Each country has intervals defined between
0.9 and 0.95 as a minimum value to apply 0.95 ≤surcharge
cos φ ≤ 1 for excessive reactive energy consumption,(9) in
SpainPFC
it is depend
0.95, thison
value
the iscurrent
set for the constraint.
legislation The maximum
in each country. Each constraint value
country is set
has to 1, because
intervals defined if
the limit is0.9
between exceeded
and 0.95theas power factor changes
a minimum valuefrom inductive
to apply to capacitive.
surcharge for excessive reactive energy
The limit in
consumption, values
Spainofit the constraints
is 0.95, can
this value is be
setadjusted to the legislation
for the constraint. in eachconstraint
The maximum country, or to the
value is
needs
set to 1,ofbecause
the installation where
if the limit the PFCMS
is exceeded the is implanted.
power factor changes from inductive to capacitive.
To define
The the population
limit values (pop), 2n can
of the constraints possible states to
be adjusted arethe
considered, in eachn country,
legislationwhere is the number of
or to the
capacitors used. Each state shows the combination
needs of the installation where the PFCMS is implanted. of turning on or off the n capacitors.
To define the population (pop), 2n possible  states are considered,
 where n is the number of
capacitors used. Each state shows the combination  X1,1 of · · turning
· X1,n on  or off the n capacitors.
 . .. 
 
pop = .. .. (10)
𝑋, ⋯ 𝑋, . . 

pop = ⋮ X2n,1 ⋱ · · ·⋮ X2n,n

(10)
𝑋 , ⋯ 𝑋 ,
Figure 1 shows TLBO flow chart.
Figure 1 shows TLBO flow chart.

Initialize number of
Select any two solution
student (populations)
randomly Xi and Xj

Calculate the mean of each

design variable If Xi better Yes
No then Xj?

Xnew=Xi+r * (Xi-Xj) Xnew=Xi+r * (Xj-Xi)

Identify the best solution
(Teacher)
Yes If new solution No
better than existing?
Modify solution based on the best
solution Xnew=Xi +r*(Xteacher- TF*Xmean)
Reject

No Yes
If new solution If new solution
better than better than existing?
existing?

Reject Accept
Final value of solution

Figure
Figure 1.
1. Flow
Flow chart
chart of
of the
the teaching
teaching learning
learning based
based optimization
optimization (TLBO)
(TLBO) Algorithm.
Algorithm.

This algorithm will be implemented in the PFCMS in real time. In addition,

addition, it will be used
used within
within
the optimization algorithm of point 2.4.

CBO Algorithm
2.4. CBO Algorithm
The power
power factor
factor compensation
compensationstudy studyofofelectrical
electricalinstallations
installations is is necessary
necessary to to perform
perform a set
a set of
of measurements
measurements of of
thethe electricalvariables
electrical variablesinvolved.
involved.Once
Oncethethemeasurements
measurementshave havebeen
been done,
done, the
capacities of the different capacitors that formed
formed the
the CB
CB can
can be
be determined.
determined.
In this
thisresearch
researcha CBO
a CBOalgorithm is proposed
algorithm to obtain
is proposed the optimal
to obtain capacitycapacity
the optimal of each ofofthe capacitors.
each of the
This algorithm
capacitors. Thisallows calculating
algorithm a capacitorsa bank
allows calculating composed
capacitors of 2 . . . n units,
bank composed achieving
of 2...n an optimal
units, achieving an
compensation. If the number of capacitors increases, a better PF compensation
optimal compensation. If the number of capacitors increases, a better PF compensation is obtained,is obtained, but it
implies
but higherhigher
it implies installation costs and
installation costsa and
morea complex control.
more complex control.
The implementation of the CBO algorithm takes place in three phases: (i) obtain the possible
solutions to cover the maximum and minimum power factor, (ii) with the set of the possible solutions,
choose those that make it possible to compensate the entire acquired field of measurements, (iii) with
the chosen solutions, select the one with the lowest capacity. Figure 2 shows CBO flow chart.
Sensors 2019, 19, 2172 7 of 23

The implementation of the CBO algorithm takes place in three phases: (i) obtain the possible
solutions to cover the maximum and minimum power factor, (ii) with the set of the possible solutions,
Sensors 2019, 19, x FOR PEER REVIEW 7 of 24
choose those that make it possible to compensate the entire acquired field of measurements, (iii) with
the chosen solutions,
Sensors 2019, select
19, x FOR PEER the one with the lowest capacity. Figure 2 shows CBO flow chart.
REVIEW 7 of 24
Initialize variables i≤n From i = 1 to n i>n From i = 1 to n
valid solutions solutions
Calculate upper and lower
Initialize variables i≤n From i = 1 to n i>n From i = 1 to n
limits PF
i≤n valid
Fromsolutions
j = 1 to n i>n solutions
TLBO minimum
i≤n Calculate upper and lower measurement data
From i =PF
1 to n i>n Capacity
limits
possible combinations i≤n From j = 1 to n i>n TLBO minimum
i≤n measurement data Capacity
From i = 1 to n i>n TLBO data(j) No If new solution
TLBO upper limit PF
possible combinations better than
existing?
No TLBO upper limit PF TLBO
Storedata(j)
result(j) No If new solution
If exist any better than
Yes
possible solution? existing?
Replace best
No Store result(j) solution
If exist any Yes
Yes
possible solution?
TLBO lower limit PF Replace best
No If all elements
solution
Yes in result are
No If exist
TLBO loweralimit
validPF valid?
No If all elements
solution? End
in result are
Yes optimization
No If exist a Yes
valid valid?
solution?
Store solution Store solution(i) End
Yes optimization
Yes
Store solution Store solution(i)

Figure
Figure 2. Flow
2. Flow chartofofthe
chart thecapacitor
capacitor bank
bankoptimization
optimization(CBO) Algorithm.
(CBO) Algorithm.

3. PFCMS Design
3. PFCMS Figure 2. Flow chart of the capacitor bank optimization (CBO) Algorithm.
Design

3.The proposed
The
PFCMS proposed
Design system
systemtakes
takes 230
230VV50 50HzHz mains supplyasasa apower
mains supply power source.
source. TheThe sampled
sampled voltage
voltage
signalssignals
and current
and currentsignals receive
signals ofofpower
receive powergrid
grid and processthrough
and process through thethe voltage
voltage sensor
sensor and the
and the
current sensor
current circuit
The proposed
sensor andand
system
circuit introduce
takes 230 in
introduce Vin the
50 Hz
themicrocontroller, that
mains supply asthat
microcontroller, performs
a power source.
performs power factor
Thefactor
power sampled calculations
voltage
calculations
usingsignals
TLBOTLBO
using and
andcurrent
switches signals receive
capacitors
and switches ofofCB.
capacitors power grid
Moreover,
of CB. andresults
Moreover, process through
are
results stored
are the
storedandvoltage
and sensorin
monitored
monitored and the using
inCDL
CDL
GS, andcurrent
using
canGS,sensor
alsoand circuit
be can and introduce
also be displayed
displayed on Android in the microcontroller,
on Android that
application.
application. performs
Figure Figure
3 shows power
3 shows factor calculations
the functional
the functional blockblock
diagram
of theusing
diagramTLBO
complete and
ofproject.switches
the complete
A smart capacitors
project.
sensor A of CB. Moreover,
smart
network sensor results
that network
allows are allows
that storedan
inspecting and monitored
inspecting
electricalan in CDL in a
electrical
installation
using GS, and
installation in can also be displayed on Android application. Figure 3 shows the functional block
non-intrusive way isa presented
non-intrusive way developing
in [33], is presented anin open-architecture
[33], developing ansmart
open-architecture
sensor network, smartbased
diagram
sensor network, based on FPGA technology, which is able to continuously monitor PQan
of the complete project. A smart sensor network that allows inspecting electrical
in industrial
on FPGA technology, which is able to continuously monitor PQ in industrial facilities, public buildings,
installation in a buildings,
facilities, public non-intrusiveand way is presented
residential homes. in [33], developing an open-architecture smart
and residential homes.
sensor network, based on FPGA technology, which is able to continuously monitor PQ in industrial
facilities, public buildings, L and residential homes.
230 V
Load
50 HzL
N
230 V
Load
50 Hz
PZEM
N (voltage and current
sensor) Capacitor bank
PZEM
(voltage and current
sensor) Capacitor bank

Arduino D1R1
USB Relay shield
(ATMEGA480)
Port
Arduino D1R1
USB Relay shield
(ATMEGA480)
Port
Data Android
Logger Aplication

Data Android
Logger Aplication
Figure 3. Power Factor Compensation and Monitoring System (PFCMS) scheme.
Figure 3. Power Factor Compensation and Monitoring System (PFCMS) scheme.

Figure 3. Power Factor Compensation and Monitoring System (PFCMS) scheme.

Sensors 2019, 19, 2172 8 of 23

3.1. Hardware

3.1.1. Microcontroller
A microcontroller is a small computer on a single integrated circuit containing a processor core,
memory, and programmable input/output peripherals [34]. An AD1R1 microcontroller (based on
ESP-8266EX) is used in this research, which has lots of libraries developed and available online for free.
Table 1 shows the main characteristics of the microcontroller.

Table 1. Arduino D1R1 characteristics [34].

Parameter Value
Microcontroller ESP-8266EX
Operating voltage 3.3 V
Input voltage 9–24 V
Digital I/O pins 11 (provide a PWM output, except D0)
Analog input pins 1
Flash memory 4 MB
Clock speed 80/160 MHz

Mains voltage and current, real power, of the network is received through the developed program
from PZEM. The reactive power and PF are calculated by the microcontroller. Moreover, the design of
switching strategy of CBs is explained in the control algorithm, in point 2.3.
Board AD1R1 allows access to Wi-Fi networks, and therefore uploading information to the Internet,
in our case to GS with times less than 1 s. It makes it ideal for the dual function of a monitoring system
and uploading information to the cloud.

3.1.2. Electric Power Meter

The PZEM model of the Ningbo Peacefair Electronic has been chosen as electrical meter, whose
characteristics are in Table 2. This equipment can measure voltage, current, active power, and active energy.
To measure the current, use a non-invasive toroidal transformer through which the phase cable passes into it.

Table 2. Peacefair PZEM characteristics [28].

Parameter Value
Voltage 80–260 V
Current 0–100 A
Active power 0–22 kW
Energy 0–9999 kWh
Measured signal Continuous wave
Communication port TTL port

Inside, the PZEM module has two opto-couplers (PC817A) has a connection with pins 1 and 2 of
the PZEM through a voltage divider with two resistors (R1 and R2) of 1 kΩ in series (extracting only
2.5 Vdc in this point). With the insertion of a resistor (R3) of 1 kΩ in parallel with the resistance R1, a
new voltage divider is obtained at that point.

3.2. Software
Two programs are developed: (i) AD1R1, which performs the functions of reading electrical
variables from PZEM, the optimization calculations and sends the data to GS; (ii) Android application,
which monitors the data stored in GS, making it possible to display industry data in real time at any
location using Smartphone or Tablet.
The measurements made with PZEM can be adjusted to different measurement times depending
on the needs of the electrical installation, this time is independent of the Internet upload time.
Sensors 2019, 19, 2172 9 of 23

Sensors 2019, 19, x FOR PEER REVIEW 9 of 24

3.2.1.Sensors
Microcontroller Program
2019, 19, x FOR PEER REVIEW 9 of 24
3.2.1. Microcontroller Program
This part of the software should be responsible for several tasks: (i) the program must start the
3.2.1.This
Microcontroller
part Programshould be responsible for several tasks: (i) the program must start the
of the software
Wi-Fi system, the electric power meter PZEM, and the initialization of the four outputs to the relays
Wi-FiThis
system,
part the
of electric
the power
software meter
bePZEM, and the
forinitialization of(i)
the four outputs to the relays
that will activate the contactors toshould
operate responsible
in output mode;several tasks:
(ii) read the the program
electrical must
variables start
from the
PZEM;
that
Wi-Fi will activate
system, the the contactors
electric power to operate
meter PZEM,in output
and the mode; (ii)
initialization read
of the
the electrical
four outputsvariables
to the from
relays
(iii) perform TLBO algorithm; (iv) data is uploaded to Internet.
PZEM;
that (iii)
will4 perform
activate theTLBO algorithm;
contactors (iv) data
to operate in is uploaded
output to Internet.
FigureFigureshows the
4 shows flow
the diagram
flow diagramcorresponding
corresponding totomode;
themain
the
(ii) read
main the electrical variables from
program.
program.
PZEM; (iii) perform TLBO algorithm; (iv) data is uploaded to Internet.
Figure 4 shows the flow diagram corresponding to the main program.
Main program

Main program
Initialize Wi-Fi
Initialize PZEM
Initialize
Initialize Wi-Fi
digitals outputs
Initialize PZEM
Initialize digitals outputs

Measure electrical variables

Measure electrical variables

Optimization TLBO algorithm

Optimization TLBO algorithm

Send data to Cloud

Data Logger
Send data to Cloud
Data Logger
Figure
Figure 4. PFCSM
4. PFCSM main
main program.
program.
Figure 4. PFCSM main program.
3.2.2.3.2.2. Android
Android AppApp
3.2.2. Android App
The Theprogramprogram developed
developed forforAndroid
Android mobile
mobile devices
devicesallows allowsvisualizing
visualizing the the
behavior of the
behavior of the
installation
The in real
program time.
developed for Android mobile devices allows visualizing the behavior of the
installation in real time.
MITAIin
installation is usedtime.
as development environment. This platform is free and available from Google for
MITAI is usedrealas development environment. This platform is free and available from Google for
the online development of apps for environment.
Android devices.
the onlineMITAI
The app
is used asof
development development
developed of
apps
with
for Android
MITAI
This platform is free and available from Google for
devices.
consists devices.
of two parts: (i) the development of screens using visual
the online development apps for Android
The app developed
components with MITAI consists
boxes,of two parts: (i)non-visual
the development of screensas usingWi-visual
The app such as buttons,
developed with lists,
MITAI text
consists and
of others,
two parts:and components
(i) the development such
of screens usingBT,
visual
components such as buttons,
Fi, etc.; (ii) programming lists,
through text boxes,
blocks thatand and
allows others,
the and and
development non-visual components suchtheas BT,
components such as buttons, lists, text boxes, others, non-visualofcomponents
the application
suchwith allWi-
as BT,
Wi-Fi,Fi,
etc.;
etc.;(ii)
(ii) programming
functionalities. through
programming through blocks
blocks thatthat allows
allows the development
the development of the application
of the application with all
with all the
Figure
the functionalities.
functionalities. 5 shows the flow diagram of the application for mobile devices.
Figure 5 shows
Figure thethe
5 shows flow diagram
flow diagramofofthetheapplication
application forformobile
mobiledevices.
devices.
Android App

Android App
Initialize
Clear variables
Initialize
Clear graphs & labels
Clear variables
Clear graphs & labels

Data change No
in Google
Data change
sheets? No
in Google
sheets?
Yes

Yes
Read n registers from Google sheets

Read n registers from Google sheets

Draw graph
Show last registers values in labels
Draw graph
Show last registers values in labels
Figure 5. PFCMS Android app flow diagram.
Figure5.5.PFCMS
Figure PFCMS Android
Android app
appflow
flowdiagram.
diagram.
 Sensors 2019, 19, 2172 10 of 23
Sensors 2019, 19, x FOR PEER REVIEW 10 of 24

3.3.Prototype
3.3. Prototype
PFCMSbuilt
PFCMS built has
has external
external connections,each
connections, each one
one connects
connects with
with the
the capacitor.The
capacitor. The communication
communication
between AD1R1 and Internet integrated inside the microcontroller, which allows data totobebe
between AD1R1 and Internet integrated inside the microcontroller, which allows data uploaded
uploaded
fromthe
from theboard,
board,without
withoutthe theneed
needforforother
otherexternal
externalcomponents.
components.
Moreover, equipment is powered by an
Moreover, equipment is powered by an external adapterexternal adapterthat
thatprovides
providesthethenecessary
necessaryvoltage
voltagetoto
AD1R1. There are other multiple ways of feeding the equipment, which can
AD1R1. There are other multiple ways of feeding the equipment, which can be implemented without be implemented without
anydifficulty
any difficultyininother
othersimilar
similarprojects.
projects.
It is possible to work with contactorswhose
It is possible to work with contactors whose coil
coil hashasa amaximum
maximum current
current ofof1010A,A,which
whichisisthe
the
one supported by the relays. In case of use contactors with coils whose current is greater than 10A, itA,
one supported by the relays. In case of use contactors with coils whose current is greater than 10
isiteasily
is easily realizable
realizable onlyonly by changing
by changing the relays
the relays to thetonext
the next
step step
of 30ofA,30 A, which
which are easily
are easily foundfound
in thein
the
market.market.
PFCMSbuilt,
PFCMS built,the
theconnection
connectionofofthethecomponents,
components, thethe control
control panel
panel and
and the
the equipment
equipmentmounted
mountedin
the container box is shown in Figure
in the container box is shown in Figure 6. 6.

(a) (b)

Figure 6. PFCMS: (a) Internal connections; (b) Final prototype.

Figure 6. PFCMS: (a) Internal connections; (b) Final prototype.
Programmer includes PFDES and capacitors values into the microprocessor. Depending on the
difference between the PFDES and power factor installation (PFINST ), the TLBO algorithm adjusts the
Programmer includes PFDES and capacitors values into the microprocessor. Depending on the
switching schemes of capacitors banks.
difference between the PFDES and power factor installation (PFINST), the TLBO algorithm adjusts the
Optimal solutions and measured electrical variables are stored in GS with the fixed measurement
switching schemes of capacitors banks.
interval. The spreadsheet represents the graphs of voltage, current, power factor, and active, reactive,
Optimal solutions and measured electrical variables are stored in GS with the fixed
and apparent powers over time.
measurement interval. The spreadsheet represents the graphs of voltage, current, power factor, and
active, reactive,
4. Results andand apparent powers over time.
Discussion
A CB
4. Results of four
and units is used to do the tests, divided into two phases: (i) obtain the optimal capacities
Discussion
for the four capacitors in each case studied, (ii) with the optimal capabilities perform the tests in real
timeA using
CB ofPFCSM.
four units is used to do the tests, divided into two phases: (i) obtain the optimal
capacities for the four capacitors in each case studied, (ii) with the optimal capabilities perform the
tests
4.1.in real
Test time using PFCSM.
Equipment

4.1.1.
4.1. TestTest Machine
Equipment
The case studies have been done in the electrical machinery laboratory of the Electrical Engineering
4.1.1. Test Machine
Department of the University of Jaen. In this laboratory, there are several trainers of electrical machines,
Thepower
in the case range
studies have
of 0.3 kWbeen done inmachine
and electric the electrical
of 1.5 kW.machinery laboratory of the Electrical
Engineering Department
This trainer of the University
has a machine of Jaen.the
that can simulate Inbehavior
this laboratory, there
as resistant are several
torque, speed trainers of
control, and
electrical
differentmachines, in the
applications power
such rangewater
as cranes, of 0.3 pumps,
kW andetc.
electric
Thismachine
machineof has1.5a kW.
control panel to allow all
This trainer has a machine that can simulate the behavior as resistant torque, speed control, and
different applications such as cranes, water pumps, etc. This machine has a control panel to allow all
Sensors 2019, 19, 2172 11 of 23

these functions. In this case, only the resistive torque mode has been used, to obtain the different PF
for each motor curves and be able to compare results.
The CB has some capacity positions and dissipate a power of up to 825 VAr.

4.1.2. Used Machines

For the studies, three types of motors have been used, two squirrel cage rotors (3-phase) of 1.5 and
0.37 kW, and a squirrel cage rotor (1-phase) with star capacitor. Table 3 illustrates the characteristics of
the machines used.

Table 3. Machine characteristics.

Squirrel Cage Squirrel Cage Squirrel Cage

Parameter
(3 Phase) (3 Phase) (1 Phase)
Rated voltage (Delta-Wye) 230/400 V 230/400 V 230 V
Rate current (Delta-Wye) 5.7/3.3 A 1/0.6 A 2.9 A
Power 1.5 kW 0.37 kW 0.37 kW
Power factor 0.76 0.83 0.96
Speed 1435 rpm 2800 rpm 2870 rpm
Frequency 50 Hz 50 Hz 50 Hz

4.1.3. Load Lamps (Electric Ballast)

For the studies, two types of lamps have been used, the sodium vapour and metal halide. Table 4
illustrate the characteristics of the lighting used.

Table 4. Lamps characteristics.

Parameter Sodium Vapour Metal Halide

Rated voltage 230 V 230 V
Rate current 1.8 A 1.15 A
Power 150 W 125 W
Frequency 50 Hz 50 Hz

In the [35] presents a comparative study of the operational characteristics of the HPS lamp when
fed by conventional ballast versus electronic ballast.
Metal halide and sodium vapour lamps have ballast with a highly non-linear load which changes
during the warm-up period.

4.1.4. Resistance-Inductance Load

The loads connected to the electrical system are mostly resistive-inductive. Moreover, resistors
and inductances have seven positions and a power of up to 1200 W and 900 VAr respectively. Table 5
shows the characteristics of these resistances and inductances.

Table 5. Resistor and Inductances characteristic.

Maximum Power Maximum Power

Position Value (Ω) Value (H)
Per Phase (VAr) Per Phase (W)
1 1050 34 4.46 46
2 750 48 3.19 65
3 435 83 1.84 110
4 300 121 1.27 160
5 213 171 0.9 230
6 150 242 0.64 330
Sensors 2019, 19, 2172 12 of 23

4.2. Capacitor Bank Optimization

The measurements carried out in the different equipment studied allow obtaining the optimum
capacity installed in the CB. The algorithm developed in 2.4 applies to the different loads to obtain the
optimum in each case.
This algorithm is developed in MatLab, since the measurements made are very extensive, and
need a longer calculation time.
The best solutions reported by this algorithm are in Table 6. As this algorithm requires a fine tuning
of controlling parameters to get the optimum results, hence the results obtained by this algorithm
may not be the global optimum solutions. However, to find out the optimum values of the common
controlling parameters such as population size and the number of iterations, the algorithm is executed a
number of times with different population sizes and number of iterations. At this point, it is important
to clarify that in the TLBO algorithm, TP updates the solution as well as in the LP.

Table 6. Comparison of optimization results of capacity for CB.

Best Solution
Capacitor Max. Optimal
Type Load Step C1 -C2 -C3 -C4
Values (µF) Possibilities Solutions
(µF)
Motor One phase 0.5 . . . 5 0.5 10,000 9931 0.5-0.5-1-2.5
Motor Three phase Delta 0.5 . . . 5 0.5 10,000 5770 0.5-0.5-0.5-0.5
Motor Three phase Wye 0.5 . . . 5 0.5 10,000 4230 0.5-0.5-0.5-0.5
Lamp Metal Halide 0.5 . . . 10 0.5 160,000 1147 5-5.5-7-8
Lamp Vapour Sodium 0.5 . . . 10 0.5 160,000 673 5-5-7-8
Lighting Parallel 0.5 . . . 10 0.5 160,000 4269 4-10-9.5-8.5
Load Resistance/Inductance Series 0.5 . . . 10 0.5 160,000 3540 0.5-1.5-2.5-2
Load Resistance/Inductance Parallel 0.5 . . . 10 0.5 160,000 9073 0.5-2-4-7.5
Daily 0.5 . . . 10 0.5 160,000 23,247 1.5-10-9.5-6.5

Table 6 shows the range of capacities, the possible solutions, the optimal solutions, and the best
solution for the time interval analysed, in each of the tests. Each test has {[max(Ci )–min(Ci )]/step}n
maximum solutions.
With the optimal solution obtained for each case of Table 6, algorithm 2.3 applies to verify that the
solution is correct in the range of measurements made.
In addition, Table 7 provides the statistical results of objective function obtained by different loads,
in which all the performance indices of TLBO are shown. Here, the initiation sets are included in
Table 6. Table 7 shows the average, standard deviation, maximum and minimum difference between
PFOBT and PFDES , of the best solution obtained for the measurements made in each case.

Table 7. Statistical results of objective function obtained by different loads con TLBO.

Standard PFOBT − PFDES

Parameter Mean
Deviation Max. Min.
Motor One phase 0.00422 0.00479 0.0174 9.3590 × 105
Motor Three phase Delta 0.00784 0.00852 0.0332 7.7156 × 105
Motor Three phase Wye 0.01669 0.01378 0.0483 7.73490 × 105
Metal Halide 0.01840 0.01058 0.0458 2.90000 × 105
Vapour Sodium 0.01775 0.01107 0.0469 1.2795 × 105
Lighting Parallel 0.00974 0.01172 0.0399 1.0196 × 104
Load Resistance/Inductance Series 0.01313 0.01158 0.0487 6.2545 × 107
Load Resistance/Inductance Parallel 0.01491 0.01625 0.050 9.6650 × 105
Daily 0.01828 0.00686 0.0285 1.3973 × 105
 Sensors 2019, 19, 2172 13 of 23
Sensors 2019, 19, x FOR PEER REVIEW 13 of 24

4.3.Compensation
4.3. CompensationPower
PowerFactor
FactorTest
TestininReal
RealTime
Time
Theenergy
The energysupply
supplycompanies
companiesincrease
increasethetheelectricity
electricitybill
billthrough
throughaasurcharge
surchargefor forexcessive
excessive
consumption of reactive energy. The usual values are surcharge below 0.95
consumption of reactive energy. The usual values are surcharge below 0.95 and without surcharge and without surcharge
between0.95
between 0.95and
and1.1.PFPFDES
DEStoto perform
perform thethe tests
tests is is 0.95.
0.95.
Different loads used are motors, lamps (ballasts),
Different loads used are motors, lamps (ballasts), and resistive and inductive
and resistive loads. This
and inductive equipment
loads. This
performs the measurements of the PFC tests and analyses the data obtained
equipment performs the measurements of the PFC tests and analyses the data obtained in real time. in real time.
PFCMShas
PFCMS hasimplemented
implementedthe thealgorithm
algorithmdescribed
describedininpoint
point2.3
2.3totoobtain
obtainthe
theoptimal
optimalcombination
combination
of capacitors installed in the
of capacitors installed in the CB. CB.
Theoptimum
The optimumcapacity
capacityisisdefined
definedby bythe
theEquations
Equations(3)‒(5),
(3)–(5),ififCB
CBisisconnected
connectedininsingle-phase,
single-phase,
three-phase,ininstar
three-phase, starorortriangle,
triangle,respectively.
respectively. InInaddition,
addition,value
valueobtained
obtainedcannot
cannotbe beadjusted
adjustedtotoaa
combination of capacitors with standard capacities existing on the market. The capacitanceobtained
combination of capacitors with standard capacities existing on the market. The capacitance obtainedis
isthe
thecapacitance
capacitanceselected
selectedininthe
theCB CBatateach
eachmoment
momentbybyusing usingTLBO.
TLBO.
Moreover, the graphs for each of the tests carried
Moreover, the graphs for each of the tests carried out show the out show thedifference
differencebetween
betweenthethecalculated
calculated
optimum capacity and the capacity obtained by TLBO in PFCMS
optimum capacity and the capacity obtained by TLBO in PFCMS to control the CB. to control the CB.

4.3.1. Motors Load

4.3.1. Motors Load
Three scenarios are defined to perform this test.
Three scenarios are defined to perform this test.
1.1. Motor
Motor0.37
0.37kW
kW(3(3phase).
phase).Delta
DeltaDDconnection.
connection.
2.2. Motor
Motor0.37
0.37kW
kW(3(3phase).
phase).Wye
WyeWWconnection.
connection.
3.3. Motor
Motor0.37
0.37kW
kW(1(1phase).
phase).
The
Thecapacity
capacityvalues
valuesused
usedfor
forTBLO
TBLOAlgorithm
Algorithmare 0.5-0.5-0.5-0.5F
are0.5-0.5-0.5-0.5 µFfor
forthe
thethree-phase
three-phasemotor
motorinin
DDand W, obtained from CBO algorithm shown
and W, obtained from CBO algorithm shown in Table 6.in Table 6.
The
Thetorque changes every
torque changes every4040s sfrom
fromnono load
load to Nm
to 0.6 0.6 Nmwithwith
a stepaof
step of 0.1with
0.1 Nm, Nm, thewith the test
test machine.
machine. The acquisition of data has been done with PZEM configured
The acquisition of data has been done with PZEM configured to measure each 1 s. to measure each 1 s.
Moreover,
Moreover, the
the test
test results of motor
results of motor D Dand
andW Wconnection,
connection,ininFigures
Figures7–9,
7–9, show
show thethe evolution
evolution of
of the
the obtained
obtained vs vs optimum
optimum capacity,
capacity, PFINST
PFINST and
and PFPF OBT, and the difference between PFOBT and PFDES, for
OBT , and the difference between PFOBT and PFDES , for
11ssapplication.
application.

(a) (b)
Figure 7. Comparison obtained vs optimal capacity: (a) Motor in W; (b) Motor in D.

Figure 7. Comparison obtained vs optimal capacity: (a) Motor in W; (b) Motor in D.

Sensors 2019,
Sensors 19, 19,
2019, x FOR PEER
x FOR REVIEW
PEER REVIEW 14 of
1424
of 24

1 2019,
Sensors
Sensors 2019,19,
19,x2172
FOR PEER REVIEW 14
14of
of24
23
1
0.95
0.98

0.9 PFINST 0.96

PFOBT
0.85 0.94
PFINST
0.92 PF
OBT
0.8
0.9
0.75
0.88
0.7
0.86
0.65 0.84

0.6 0.82
0 50 100 150 200 0 50 100 150 200
Time (s) Time (s)

(a)(a) (b)(b)

Figure 8. Comparison (a)power factor installation (PFINST) vs power factor obtained

of (b) (PFOBT): (a) Motor
Figure 8. Comparison of power factor installation (PFINST) vs power factor obtained (PFOBT): (a) Motor
in W;
in (b)(b)
W; Motor in D.
Motor in D.
Figure 8. Comparison of power factor installation (PFINST ) vs power factor obtained (PFOBT ): (a) Motor
in W; (b)
Figure Motor in D. of power factor installation (PFINST) vs power factor obtained (PFOBT): (a) Motor
8. Comparison
in W;0.05
(b) Motor in D.
Motor in W
0.045 Motor in D

0.04

0.035
PFOBT - PFDES

0.03

0.025

0.02

0.015

0.01

0.005

0
0 50 100 150 200
Time (s)

Figure 9. Difference
Figure
Figure 9. between
9. Difference
Difference PFOBT
between
between PF
PF and power
and
and
OBT
OBT
factor
power
power desired
factor
factor (PF(PF
DES).(PF
desired
desired DES). DES ).

Figure 1010
Figure
Figure illustrates
10 thethe
illustrates
illustrates assembly
the done
assembly
assembly in in
done
done thethe
in laboratory.
the laboratory.
laboratory.
Figure 9. Difference between PFOBT and power factor desired (PFDES).

Figure 10 illustrates the assembly done in the laboratory.

Figure 10.10.
Figure Final assembly
Final forfor
assembly thethe
study of the
study three
of the phase
three motor
phase load.
motor load.

Figure 10. Final assembly for the study of the three phase motor load.
 Sensors 2019, 19, x FOR PEER REVIEW 15 of 24

Sensors 2019, 19, 2172 15 of 23

The capacity values used for TBLO algorithm are 0.5-0.5-1-2.5 F for the single-phase motor,
Sensors 2019, 19, x FOR PEER REVIEW 15 of 24
obtained with CBO algorithm are in Table 6.
The
The
The test results
capacity
capacity of motor
values
values usedone-phase,
used forTBLO
for shown
TBLOalgorithm in Figures
algorithm 11 and 12,F
are0.5-0.5-1-2.5
are 0.5-0.5-1-2.5 are
µFforthe evolution
forthe of the obtained
thesingle-phase
single-phase motor,
motor,
vs optimum
obtained with capacity,
CBO the
algorithmPF are
INST
obtained with CBO algorithm are in Table 6.and
in power
Table 6. factor obtained (PFOBT), and the difference between PFOBT

andThe
PFtest
TheDES , for
test 1 s of
results
results TLBO
ofof
motor algorithm
motor one-phase,
one-phase, application. Figure
shownininFigures
shown 11a
Figures11 shows
11and
and aare
12,are
12, PF the
the greater than
INST evolution
evolution 0.95,
ofofthe
the with an
obtained
obtained
optimal
vsvsoptimumcapacity
optimumcapacity, less
capacity, thethan
the PF zero
PFINST and the
andpower
INSTand
obtained
powerfactor capacity
factorobtained
obtained(PFequal
(PFOBT to zero.
),),and
OBT andthe
thedifference
differencebetween
betweenPF PFOBT
OBT
andPF
and PFDES
DES , for
, for 1 s1 of
s of TLBOalgorithm
TLBO algorithmapplication.
application.Figure
Figure11a
11ashows
showsaaPF PFINST
INST greaterthan
greater than0.95,
0.95,with
withan
an
optimalcapacity
optimal capacityless lessthan
thanzero
zeroand
andthe
theobtained
obtainedcapacity
capacityequal
equaltotozero.
zero.

(a) (b)

(a) (b)
Figure 11. One-phase motor comparison: (a) Obtained vs optimal capacity; (b) PFINST vs PFOBT.
Figure 11. One-phase motor comparison: (a) Obtained vs optimal capacity; (b) PFINST vs PFOBT .
Figure 11. One-phase motor comparison: (a) Obtained vs optimal capacity; (b) PFINST vs PFOBT.

Figure 12. Difference

Figure between
12. Difference PFOBT
between PFand PFDES
OBT and (1-phase
PFDES motor).
(1-phase motor).

The
Thedifference
differencebetween
betweenPF PFOBT and PF DES,, for
OBT and PFDES
for three-phase
three-phase motors
motors in
in D
D and
and W,
W,isislower
lowerthan
than0.0332
0.0332
and Figure 12. Difference between PF OBT and PFDES (1-phase motor).
and0.0483
0.0483respectively,
respectively,andandfor
forthethesingle-phase
single-phasemotormotorititisislower
lowerthan
than0.0174.
0.0174.The
Theresults
resultsobtained
obtained
indicate
indicate that
thatTLBO
TLBO algorithm
algorithm isisapplicable
applicable totothis
thistype
typeofofproblems.
problems.
The difference between PFOBT and PFDES, for three-phase motors in D and W, is lower than 0.0332
and 0.0483 respectively,
4.3.2. Lamps Load and for the single-phase motor it is lower than 0.0174. The results obtained
4.3.2. Lamps Load
indicate that TLBO algorithm is applicable to this type of problems.
The facilities use a variety of lamps for lighting. This test utilizes sodium vapour and metal halide
The facilities use a variety of lamps for lighting. This test utilizes sodium vapour and metal
lamps.
4.3.2. In order
Lamps Load to perform this test, it defines two scenarios.
halide lamps. In order to perform this test, it defines two scenarios.
1.1. The facilities
Metal
Metal halideuse
halide a variety of lamps for lighting. This test utilizes sodium vapour and metal
lamp.
lamp.
halide lamps.
2.2. Sodium In order
Sodium vapour to perform this test, it defines two scenarios.
vapour lamp.
lamp.
1. MetalThehalide lamp. the process of starting the lamps up to the permanent operating mode. The
tests illustrate
illustrate
The tests the process of starting the lamps up to the permanent operating mode.
2. starting
Sodium vapour
time is lamp.2 min. Moreover, graphs show 240 s of operation, the permanent regime reach
approx.
The starting time is approx. 2 min. Moreover, graphs show 240 s of operation, the permanent regime
The tests illustrate the process of starting the lamps up to the permanent operating mode. The
starting time is approx. 2 min. Moreover, graphs show 240 s of operation, the permanent regime reach
Sensors 2019, 19, x FOR PEER REVIEW 16 of 24

to 120 s. The capacity values used for TBLO algorithm are 5-5.5-7-8 and 5-5-7-8 F for metal halide
Sensors 2019, 19, 2172 16 of 23
Sensors
and 2019, 19,vapour
sodium x FOR PEER REVIEW obtained from CBO algorithm show in Table 6.
respectively, 16 of 24

Figures 13‒15 show the evolution of the obtained vs optimum capacity, PFINST and PFOBT, and the
to 120 s. The capacity values used for TBLO algorithm are 5-5.5-7-8 and 5-5-7-8 F for metal halide
difference
reach to 120 between
s. ThePF OBT andvalues
capacity PFDES for
useddifferent
for TBLOlamps, with an
algorithm areapplication time
5-5.5-7-8 and of 1 µF
5-5-7-8 s for
forTLBO
metal
and sodium vapour respectively, obtained from CBO algorithm show in Table 6.
algorithm.
halide and sodium vapour respectively, obtained from CBO algorithm show in Table 6.
Figures 13‒15 show the evolution of the obtained vs optimum capacity, PFINST and PFOBT, and the
Figures 13–15 show the evolution of the obtained vs optimum capacity, PFINST and PFOBT , and
difference between PFOBT and PFDES for different lamps, with an application time of 1 s for TLBO
the difference between PFOBT and PFDES for different lamps, with an application time of 1 s for
algorithm.
TLBO algorithm.

(a) (b)

(a)
Figure 13. Comparison (b)Sodium vapour.
obtained vs optimal capacity: (a) Metal halide; (b)
Figure 13. Comparison obtained vs optimal capacity: (a) Metal halide; (b) Sodium vapour.

Figure 13. Comparison obtained vs optimal capacity: (a) Metal halide; (b) Sodium vapour.

(a) (b)
Sensors 2019, 19, x FOR PEER REVIEW 17 of 24
Figure 14. Comparison PFINST vs PFOBT : (a) Metal halide; (b) Sodium vapour.
(a)Comparison PFINST vs PFOBT: (a) Metal halide; (b) Sodium
Figure 14. (b) vapour.

Figure 14. Comparison PFINST vs PFOBT: (a) Metal halide; (b) Sodium vapour.

Figure Difference
15.15.
Figure between
Difference PFPF
between OBT and PFDES
OBT and PF
(Metal halide).
DES (Metal halide).

The difference between PFOBT and PFDES, for sodium vapour and metal halide lamps, is less than
0.0458 and 0.0469 respectively. The results obtained indicate that the TLBO algorithm is applicable to
lighting installations.
Once the assembly is done in the laboratory, it remains in the shape shown in Figure 16.
 0.01

0.005

0
0 50 100 150 200
Time (s)
Sensors 2019, 19, 2172 17 of 23
Figure 15. Difference between PFOBT and PFDES (Metal halide).

The
The difference
difference between
between PFPF OBT
OBT and and
PFPF , for, for
DESDES sodium
sodium vapour
vapour andand metal
metal halide
halide lamps,
lamps, is less
is less than
than
0.0458 and 0.0469 respectively. The results obtained indicate that the TLBO algorithm is applicable to to
0.0458 and 0.0469 respectively. The results obtained indicate that the TLBO algorithm is applicable
lighting
lighting installations.
installations.
Once thethe
Once assembly
assembly is done
is done in in
thethe laboratory,
laboratory, it it remains
remains inin the
the shape
shape shown
shown ininFigure
Figure16.16.

Figure 16.
Figure 16. Final
Final assembly for the
assembly for the study
study of
of the
the lamps
lamps load.
load. Sodium
Sodium vapour.
vapour.

4.3.3.
4.3.3. Resistance
Resistance andand Inductance
Inductance Load
Load
In In order
order to to
dodo this
this test,
test, it defines
it defines twotwo scenarios,
scenarios, in in which
which thethe load
load was
was taken
taken resistance
resistance fixed
fixed and
and
thethe inductance
inductance change.
change. In In this
this case,
case, thethe defined
defined scenarios
scenarios areare
as as follows:
follows:
1. 1. Resistance-Inductance series
Resistance-Inductance loadload
series
2. 2. Resistance-Inductance parallel
Resistance-Inductance loadload
parallel
Tests show
Tests show thethe
different operating
different states
operating of theofloads.
states The reactance
the loads. variation
The reactance time is 40
variation seconds
time is 40 s
approximately. Moreover, capacity values used are 0.5-2-4-7.5 and 0.5-1.5-2.5-2 μF,
approximately. Moreover, capacity values used are 0.5-2-4-7.5 and 0.5-1.5-2.5-2 µF, for parallelfor parallel andand
series respectively, obtained from the CBO algorithm show in
series respectively, obtained from the CBO algorithm show in Table 6. Table 6.
The test
The results
test reflect
results inin
reflect Figure 17‒19,
Figures thethe
17–19, evolution
evolutionof of
thethe
obtained
obtainedvsvs
optimum
optimum capacity,
capacity,PF
PFINST
INST
and PF OBT, and the difference between PFOBT and PFDES, with an application time of 1 s for the TLBO
and PFOBT , and the difference between PFOBT and PFDES , with an application time of 1 s for the
algorithm.
Sensors
TLBO 2019, 19, x FOR PEER REVIEW
algorithm. 18 of 24

(a) (b)

Figure 17. Comparison obtained vs optimal capacity: (a) Series connection; (b) Parallel connection.

Figure 17. Comparison obtained vs optimal capacity: (a) Series connection; (b) Parallel connection.
 (a) (b)

(a) (b)
Figure 17. Comparison obtained vs optimal capacity: (a) Series connection; (b) Parallel connection.
Sensors 2019, 19, 2172 18 of 23
Figure 17. Comparison obtained vs optimal capacity: (a) Series connection; (b) Parallel connection.

(a) (b)

(a) (b)
Figure 18. Comparison PFINST vs PFOBT: (a) Series connection; (b) Parallel connection.
Figure 18. Comparison PFINST vs PFOBT : (a) Series connection; (b) Parallel connection.

Figure 18. Comparison PFINST vs PFOBT: (a) Series connection; (b) Parallel connection.

Figure Difference
19.19.
Figure between
Difference PFPF
between OBTand
OBT andPF DES.
PFDES

The
Thedifference between
difference PF
between and
PFOBT
OBT
Figure
PFDES
19.and
, for
PFDES
Difference
series
, between
for and
series parallel
PFand
loads, is less than 0.050. The results
parallel
OBT and PFDES. loads, is less than 0.050. The
obtained with thewith
results obtained TLBO algorithm
the indicate indicate
TLBO algorithm the correcttheoperation for different
correct operation loads. loads.
for different
The difference between PFOBT and PFDES, for series and parallel loads, is less than 0.050. The
4.3.4. Daily Use of IM
results obtained with the TLBO algorithm indicate the correct operation for different loads.
Data used correspond to one day (86,400 s) of 1.5 kW motor operation. The capacity values used
for TBLO algorithm are 1.5-10-9.5-6.5, obtained from the CBO algorithm show in Table 6.
The result of the measure data is reflected in the three graphs, Figures 20–22, the evolution of
the obtained vs optimum capacity, PFINST and PFOBT , and the difference between PFOBT and PFDES,
applying the TLBO algorithm every second.
Figure 21 shows a PFINST greater than 0.95 in some time intervals, with an optimal capacity less
than zero and the obtained capacity equal zero, as shown in Figure 20.
The difference between PFOBT and PFDES , for induction motors, is less than 0.03. The results
obtained with the TLBO algorithm are applicable to these engines.
 for TBLO algorithm are 1.5-10-9.5-6.5, obtained from the CBO algorithm show in Table 6.
obtained vs optimum capacity, PFINST and PFOBT, and the difference between PFOBT and PFDES, applying
The result of the measure data is reflected in the three graphs, Figure 20–22, the evolution of the
the TLBO algorithm every second.
obtained vs optimum capacity, PFINST and PFOBT, and the difference between PFOBT and PFDES, applying
Figure 21 shows a PFINST greater than 0.95 in some time intervals, with an optimal capacity less
the TLBO algorithm every second.
thanSensors
zero2019,
and the obtained capacity equal zero, as shown in Figure 20.
Figure 19,
212172 19 of 23
shows a PFINST greater than 0.95 in some time intervals, with an optimal capacity less
than zero and the obtained capacity equal zero, as shown in Figure 20.

Figure 20. Comparison obtained vs optimal capacity of daily.

Figure Comparison
20. 20.
Figure obtained
Comparison vs optimal
obtained capacity
vs optimal of daily.of daily.
capacity

Sensors 2019, 19, x FOR PEER REVIEW 20 of 24

Figure 21. Comparison PFINST vs PFOBT of daily.
Figure 21. Comparison PFINST vs PFOBT of daily.

Figure 21. Comparison PFINST vs PFOBT of daily.

Figure22.
Figure Difference between
22. Difference betweenPFPF and PFDES
OBT and PF
OBT of daily.
DES of daily.

The difference between PFOBT and PFDES, for induction motors, is less than 0.03. The results
obtained with the TLBO algorithm are applicable to these engines.

4.4. Cloud Data Logger

 0 1 2 3 4 5 6 7 8
Time (s) 104

Figure 22. Difference between PFOBT and PFDES of daily.

The difference between PFOBT and PFDES, for induction motors, is less than 0.03. The results
Sensors 2019, 19, 2172 20 of 23
obtained with the TLBO algorithm are applicable to these engines.

4.4.
4.4. Cloud
CloudData
DataLogger
Logger
The
The previous
previous tests
tests reflect differenttime
reflect different timeintervals,
intervals,but
butininorder
ordertotoobtain
obtainthe thecomplete
complete monitored
monitored of
of
thethe industry,
industry, it necessary
it is is necessaryto to data
data stored
stored during
during thethe whole
whole period
period of of operation
operation of of
thethe equipment.
equipment.
As
Asanan example,
example,this thissection
sectionpresents
presents the
the evolution
evolutionofof the
the operation
operationof of the
the 1.5
1.5 kW
kW motor
motor during
during
one week.
one week.
The
Thedata
datastores
storesevery
everysecond,
second,which
whichcoincides with
coincides thethe
with timetime
of completion
of completion of theofmeasures. It is
the measures.
possible to modify the measurement interval depending on the monitoring
It is possible to modify the measurement interval depending on the monitoring needs. Also, data isneeds. Also, data is
available
available to
to download
download and and visualize
visualize in
in the
the desired
desired time
time interval.
interval.
Figures
Figure 2323show
showthe thegraphs
graphsin inreal
realtime
timedepending
dependingon onthethedata
datauploaded
uploadedduring
duringaaweekweekininGS.
GS.

Figure
Figure 23.
23. Graphs
Graphssheet.
sheet.

4.5. App
4.5. Appin
inReal
RealTime
Time
Finally, itit has
Finally, has been
been tested
tested the
the operation
operationmaking
makinguse
useofofthe
theapps
appsdeveloped
developedfor
forAndroid
Androidsystems.
systems.
The application developed makes it possible to monitor the installation via Wi-Fi or 4G
The application developed makes it possible to monitor the installation via Wi-Fi or 4G connectionconnection
The app shows the last 160 values measured by compensation power factor, therefore, taking
into account that the measurements are made every second, the last 160 measurements made that
will correspond to the last 160 s of operation. In addition to the graphs, the spreadsheet shows the
measured instant data
The test was done on a Smartphone and Tablet. Following are screenshots of each of the capacity
needed and calculated to different loads, shown in Figure 24.
 The app shows the last 160 values measured by compensation power factor, therefore, taking
into account that the measurements are made every second, the last 160 measurements made that
will correspond to the last 160 seconds of operation. In addition to the graphs, the spreadsheet shows
the measured instant data
The test was done on a Smartphone and Tablet. Following are screenshots of each of the capacity
Sensors 2019, 19, 2172 21 of 23
needed and calculated to different loads, shown in Figure 24.

(a) (b)

Figure 24. Permanent regime: (a) Lamps metal halide; (b) Resistance and Inductance Parallel.
Figure 24. Permanent regime: (a) Lamps metal halide; (b) Resistance and Inductance Parallel.
5. Conclusions
This project performs the design and implements the compensation power factor using TLBO,
5. Conclusion
determines the best solution by CBO algorithm, and monitors the system in real-time using CDL.
This project performs the design and implements the compensation power factor using TLBO,
Results obtained with the TLBO algorithm, allow us to reduce the processing time of the measured
determines the best solution by CBO algorithm, and monitors the system in real-time using CDL.
data to achieve the best solution, that determines the switching of capacitors in order to compensate
Results obtained with the TLBO algorithm, allow us to reduce the processing time of the measured
demand of excessive reactive power, thus bringing the power factor near to desired level by means of
data to achieve the best solution, that determines the switching of capacitors in order to compensate
the CBO algorithm, the values of the optimal capacitors to obtain and install. The data stored in the
demand of excessive reactive power, thus bringing the power factor near to desired level by means
cloud enables the analysis and improvement of the operation of the installation.
of the CBO algorithm, the values of the optimal capacitors to obtain and install. The data stored in
The model of PFCMS shown in this paper allows significant economic savings, both in equipment
the cloud enables the analysis and improvement of the operation of the installation.
cost and wiring. That allows visualizing the electrical variables without needing to be present in the
The model of PFCMS shown in this paper allows significant economic savings, both in
place where the measurements are being made.
equipment cost and wiring. That allows visualizing the electrical variables without needing to be
present in the
Author place where All
Contributions: the authors
measurements are beingactively
have contributed made. and fundamentally to the development of the
presented work. A.C.O. and F.J.S.S. have developed the hardware and software design, in addition to the
Author Contributions:
programming All authorsand
the prototype have
thecontributed
Android app. actively and fundamentally
The writing of the papertohas
thebeen
development of the
done by each author
attending
presented work.to Antonio
their corresponding
Cano Ortega part
andof Francisco
the development of the work.
José Sánchez Sutil have developed the hardware and
software design,This
Funding: in addition
research to the programming
received no external the prototype and the Android app. The writing of the paper
funding.
has been done by each author attending to their corresponding part of the development of the work.
Acknowledgments: The authors would like to thank the Department of Electrical Engineering of the University
of Jaen,
Funding: Thisfor allowing
research the useno
received of external
their laboratories
funding. and material in the development of this research.
Conflicts of Interest: The authors declare no conflict of interest.
Acknowledgments: The authors would like to thank the Department of Electrical Engineering of the University
of Jaen, for allowing the use of their laboratories and material in the development of this research.
Abbreviations
Conflicts of Interest:
The following The authors use
abbreviations declare nomanuscript:
in this conflict of interest.

Abbreviations
AD1R1 Arduino D1R1
CB Capacitor Bank
CBO Capacitor Bank Optimization
CDL Cloud Data Logger
Sensors 2019, 19, 2172 22 of 23

DL Data Logger
GS Google Sheets
IM Induction Motor
MITAI MIT App Inventor 2
P Active Power
PF Power Factor
PFDES Power Factor Desired
PFINST Power Factor Installation
PFOBT Power Factor Obtained
PFC Power Factor Correction
PFCMS Power Factor Compensation and Monitoring System
PLC Programmable Logic Controller
pop Population
PZEM PZEM-004t
Q Reactive Power
TLBO Teaching Learning Based optimization
SVC Static Var Compensator

References
1. Rao, R.V.; Savsani, V.J.; Vakharia, D.P. Teaching–learning-based optimization: A novel method for constrained
mechanical design optimization problems. Comput. Aided Des. 2011, 43, 303–315. [CrossRef]
2. Rao, R.V.; Vakharia, D.P. Parameter optimization of modern machining processes using teaching–learning-based
optimization algorithm. Eng. Appl. Artif. Intell. 2013, 26, 524–531.
3. Nayaka, M.R.; Nayakb, C.K.; Routc, P.K. Application of Multi-Objective Teaching Learning Based
Optimization Algorithm to Optimal Power Flow Problem. Procedia Technol. 2012, 6, 255–264. [CrossRef]
4. Mohanty, B.; Tripathy, S. A teaching learning based optimization technique for optimal location and size of
DG in distribution network. J. Electr. Syst. Inf. Technol. 2016, 3, 33–44. [CrossRef]
5. Sultana, S.; Kumar Roy, P. Optimal capacitor placement in radial distribution systems using teaching learning
based optimization. Electr. Power Energy Syst. 2014, 54, 387–398. [CrossRef]
6. Bhattacharyya, B.; Babu, R. Teaching Learning Based Optimization algorithm for reactive power planning.
Electr. Power Energy Syst. 2016, 81, 248–253. [CrossRef]
7. Dixon, J.; Morán, L.; Rodríguez, J.; Domke, R. Reactive power compensation technologies: State-of-the-art
review. Proc. IEEE 2005, 93, 2144–2164. [CrossRef]
8. García-Canseco, E.; Griñó, R.; Ortega, R.; Salichs, M.; Stankovic, A.M. Power-factor compensation of electrical
circuits. IEEE Control Syst. Mag. 2007, 27, 46–59.
9. Gayatri, M.T.L.; Parimi, A.M.; Pavan Kumar, A.V. A review of reactive power compensation techniques in
microgrids. Renew. Sustain. Energy Rev. 2018, 81, 1030–1036. [CrossRef]
10. Zheng, F.; Zhang, W. Long Term Effect of Power Factor Correction on the Industrial Load: A Case Study. In
Proceedings of the IEEE Australasian Universities Power Engineering Conference (AUPEC), Melbourne,
VIC, Australia, 19–22 November 2017.
11. Zhang, W.; Feng, G.; Liu, Y.; Wu, B. New Digital Control Method for Power Factor Correction. IEEE Trans.
Ind. Electron. 2006, 53, 3. [CrossRef]
12. Colak, I.; Bayindir, R.; Bay, F. Reactive power compensation using a fuzzy logic controlled synchronous
motor. Energy Convers. Manag. 2003, 442, 2189–2204.
13. Jain, R.; Sharma, S.; Sreejeth, M.; Singh, M. PLC based Power Factor Correction of 3-Phase Induction Motor.
In Proceedings of the IEEE International Conference on Power Electronics, Intelligent Control and Energy
Systems, Delhi, India, 4–6 July 2016.
14. Kulkarni, N.; Vijay Murthy, H.; Raju, A. PLC based intelligent power factor correctors for industrial power
systems-A case study. In Proceedings of the International Conference on Power and Advanced Control
Engineering, Bengaluru, India, 12–14 August 2015.
15. Guillardi Júniora, H.; Verri Liberadob, E.; Antenor Pomilioa, J.; Pinhabel Marafãob, F. General-compensation-
purpose Static Var Compensator Prototype. Hardware X 2018, 5, e00049. [CrossRef]
Sensors 2019, 19, 2172 23 of 23

16. Kabir, Y.; Mohammad Mohsin, Y.; Monirujjaman Khan, M. Automated Power Factor Correction and Energy
Monitoring System. In Proceedings of the IEEE International Conference on Electrical, Computer and
Communication Technologies (ICECCT), Coimbatore, India, 22–24 February 2017.
17. Tubburee, O.; Kumsuwan, Y. Current Sector Control for a Three-Phase Three-Level PFC Rectifier. In
Proceedings of the 9th International Conference on Power Electronics and ECCE Asia (ICPE-ECCE Asia),
Seoul, Korea, 1–5 June 2015.
18. Pothirasan, N.; Pallikonda Rajasekaran, M.; Muneeswaran, V. Real time reactive power compensation for
battery/photovoltaic hybrid power source for internet of hybrid electric vehicle system. Cogn. Syst. Res.
2018, 52, 473–488. [CrossRef]
19. Roman-Loera, A.; Flores, L.A.; Rizo-Diaz, F.D.J.; Arambula-Miranda, L.E. Current Sensorless Predictive
Algorithm Control for Three-Phase Power Factor Correction. In Proceedings of the 34th Annual Conference
of IEEE Industrial Electronics, Orlando, FL, USA, 10–13 November 2008.
20. IEEE Standards 18-2012. IEEE Standard for Shunt Power Capacitors; Institute of Electrical and Electronics
Engineers, Inc.: Piscataway, NJ, USA, 2012.
21. C37.26-2014-IEEE. Guide for Methods of Power-Factor Measurement for Low-Voltage (1000 V AC or lower) Inductive
Test Circuits; Institute of Electrical and Electronics Engineers, Inc.: Piscataway, NJ, USA, 2014.
22. C37.99-2012-IEEE. Guide for the Protection of Shunt Capacitor Banks; Institute of Electrical and Electronics
Engineers, Inc.: Piscataway, NJ, USA, 2012.
23. Jabbar Mnati, M.; Van den Bossche, A.; Farhood Chisab, R. A Smart Voltage and Current Monitoring System
for Three Phase Inverters Using an Android Smartphone Application. Sensors 2017, 17, 872. [CrossRef]
[PubMed]
24. Lin, C.M.; Chen, M.T. Design and Implementation of a Smart Home Energy Saving System with Active
Loading Feature Identification and Power Management. In Proceedings of the IEEE 3rd International Future
Energy Electronics Conference and ECCE Asia (IFEEC 2017-ECCE Asia), Kaohsiung, Taiwan, 3–7 June 2017.
25. Martin Tureczek, A.; Sieverts Nielsen, P. Structured literature review of electricity consumption classification
using smart meter data. Energies 2017, 10, 584. [CrossRef]
26. Mit App Inventor 2. Available online: https://ai2.appinventor.mit.edu (accessed on 25 March 2019).
27. Ningbo Peacefair Electronic Co., Ltd. Available online: https://peacefair.en.made-in-china.com (accessed on
25 March 2019).
28. Abul’Wafa, A.R. Optimal capacitor placement for enhancing voltage stability in distribution systems using
analytical algorithm and Fuzzy-Real Coded GA. Electr. Power Energy Syst. 2014, 55, 246–252. [CrossRef]
29. El-Fergany, A.A.; Abdelaziz, A.Y. Capacitor placement for net saving maximization and system stability
enhancement in distribution networks using artificial bee colony-based approach. Electr. Power Energy Syst.
2014, 54, 235–243. [CrossRef]
30. Kolenc, M.; Papic, I.; Blazic, B. Coordinated reactive power control to achieve minimal operating costs.
Electr. Power Energy Syst. 2014, 63, 1000–1007. [CrossRef]
31. Homaee, O.; Zakariazadeh, A.; Jadid, S. Real-time voltage control algorithm with switched capacitors in
smart distribution system in presence of renewable generations. Electr. Power Energy Syst. 2014, 54, 187–197.
[CrossRef]
32. Rao, R.V.; Savsani, V.J.; Vakharia, D.P. Teaching–Learning-Based Optimization: An optimization method for
continuous non-linear large scale problems. Inf. Sci. 2012, 183, 1–15. [CrossRef]
33. Morales-Velazquez, L.; Romero-Troncoso, R.D.; Herrera-Ruiz, G.; Morinigo-Sotelo, D.; Osornio-Rios, R.A.
Smart sensor network for power quality monitoring in electrical installations. Measurement 2017, 103, 133–142.
[CrossRef]
34. Wemos Electronics. Available online: https://wiki.wemos.cc/products:d1:d1 (accessed on 25 March 2019).
35. Ahmed, A.M.; Osama, A.A. Comparative study of 250 W high pressure sodium lamp operating from both
conventional and electronic ballast. J. Electr. Syst. Inf. Technol. 2014, 1, 234–254.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).