sensors

Article
Power Quality Disturbance Tracking Based on a Proprietary
FPGA Sensor with GPS Synchronization
Oscar N. Pardo-Zamora 1 , Rene de J. Romero-Troncoso 1 , Jesus R. Millan-Almaraz 2 ,
Daniel Morinigo-Sotelo 3 , Roque A. Osornio-Rios 1 and Jose A. Antonino-Daviu 4, *

1 Faculty of Engineering, Autonomous University of Queretaro, San Juan del Rio 76806, Mexico;
opardo10@alumnos.uaq.mx (O.N.P.-Z.); troncoso@hspdigital.org (R.d.J.R.-T.);
raosornio@hspdigital.org (R.A.O.-R.)
2 Faculty of Physical, and Mathematical Sciences, Autonomous University of Sinaloa, Culiacan 80040, Mexico;
jrmillan@uas.edu.mx
3 Department of Electric Engineering, University of Valladolid, 47011 Valladolid, Spain;
daniel.morinigo@eii.uva.es
4 Instituto Tecnológico de la Energía, Universitat Politecnica de Valencia, 46022 Valencia, Spain
* Correspondence: joanda@die.upv.es

Abstract: The study of power quality (PQ) has gained relevance over the years due to the increase in
non-linear loads connected to the grid. Therefore, it is important to study the propagation of power
quality disturbances (PQDs) to determine the propagation points in the grid, and their source of



generation. Some papers in the state of the art perform the analysis of punctual measurements of


a limited number of PQDs, some of them using high-cost commercial equipment. The proposed
Citation: Pardo-Zamora, O.N.; method is based upon a developed proprietary system, composed of a data logger FPGA with GPS,
Romero-Troncoso, R.d.J; that allows the performance of synchronized measurements merged with the full parameterized
Millan-Almaraz, J.R.; PQD model, allowing the detection and tracking of disturbances propagating through the grid using
Morinigo-Sotelo, D.; Osornio-Rios, wavelet transform (WT), fast Fourier transform (FFT), Hilbert–Huang transform (HHT), genetic
R.A.; Antonino-Daviu, J.A. Power
algorithms (GAs), and particle swarm optimization (PSO). Measurements have been performed in an
Quality Disturbance Tracking Based
industrial installation, detecting the propagation of three PQDs: impulsive transients propagated at
on a Proprietary FPGA Sensor with
two locations in the grid, voltage fluctuation, and harmonic content propagated to all the locations.
GPS Synchronization. Sensors 2021,
21, 3910. https://doi.org/10.3390/
The results obtained show that the low-cost system and the developed methodology allow the
s21113910 detection of several PQDs, and track their propagation within a grid with 100% accuracy.

Academic Editor: Yolanda Vidal Keywords: global positioning system; industrial facilities; propagation; power quality disturbance;
particle swarm optimization; genetic algorithms; field-programmable gate array
Received: 11 May 2021
Accepted: 3 June 2021
Published: 5 June 2021
1. Introduction
Publisher’s Note: MDPI stays neutral Nowadays, power quality (PQ) is a combination of characteristics and conditions of
with regard to jurisdictional claims in
the power supplied to the equipment to guarantee its continuous operation, and its studies
published maps and institutional affil-
are important for industrial processes to maintain the quality standards of the power grid
iations.
and to avoid damage to equipment connected to the grid [1]. Disturbances cause a poor PQ
and are produced by non-linear loads connected to the grid; these sources can be connected
at long distances from the point of interest to be studied. The propagation of disturbances
strongly depends on the topology of each grid and the impedance, so disturbances can
Copyright: © 2021 by the authors. decay with distance from their point of origin, but they can also be amplified [2]. There
Licensee MDPI, Basel, Switzerland. are commercial devices that can measure and analyze electrical signals; for example, the
This article is an open access article
SEL-735 PQ and Revenue Meter [3] is a modular PQ meter which allows the capture
distributed under the terms and
of power quality disturbances (PQDs), but does not allow synchronized measurements
conditions of the Creative Commons
to monitor the propagation of disturbances, and it is expensive. The Fluke 1760TR PQ
Attribution (CC BY) license (https://
Analyzer [4] is a single point measurement device that can synchronize measurements
creativecommons.org/licenses/by/
with other devices, but it can only measure certain PQ parameters for short periods of
4.0/).

Sensors 2021, 21, 3910. https://doi.org/10.3390/s21113910 https://www.mdpi.com/journal/sensors

Sensors 2021, 21, 3910 2 of 21

time, so it cannot perform continuous PQ monitoring, nor is it suitable for monitoring

the propagation of disturbances because it requires the acquisition of several devices,
and it is a high-cost device. Due to the mentioned limitations of commercial equipment,
it is important to develop a system that detects all PQDs and performs synchronized
measurements to monitor the propagation of disturbances, and is cost-effective.
Aiming to analyze PQDs, different signal processing techniques have been developed
over the years to perform the detection, classification, and quantification of PQDs using
different mathematical models developed for each one depending on its application. The
authors in [5] performed a compilation of detection and classification techniques for PQDs
generated by renewable energy sources in a grid. This work provides knowledge of the
latest techniques developed for PQ diagnosis. The authors in [6] provided a comprehensive
review on the state-of-art techniques based on digital signal processing and machine learn-
ing for the automatic recognition of PQ events. The authors in [7] developed a structured
methodology in combination with a mathematical model which describes waveforms
that contain simultaneous PQDs capable of being adjusted to reproduce PQDs contained
in electric waveforms. This model was tested with recorded signals, and proved to be
able to reproduce the signals with minimal error. Nevertheless, the extraction process
is semi-automatic, and requires the support of other techniques for this purpose. The
authors in [8] introduced a wavelet-based PQ indicator. An instantaneous disturbance
index (ITD) and global disturbance ratio index (GDR) are defined to integrally reflect
the PQ level in the power distribution network under steady-state and/or transient con-
ditions. The effectiveness of this method has been proved by comparing the proposed
PQ indicators with classical indices, and the results confirm that the method efficiently
extracts the characteristics of each component from the multi-event test signal, but it has
not been implemented in hardware. In [9], the authors submitted a method of detection and
classification of PQDs using different wavelets and neural network classifiers. Different
wavelets were used to extract features of the raw signals, and a neural network was used to
detect the PQDs. Simulation results showed the performance of the network for different
wavelets and its efficacy, but improved accuracy can be obtained if the Fuzzy Technique is
employed for the detection of PQDs. Some detection and classifications techniques have
been implemented in PQ measurement systems to detect fault events in real time, whereas
the authors in [10] proposed a smart sensor that allows the detection, classification, and
quantification of PQDs, where it uses Hilbert transform techniques for detection, a feed
forward neuronal network (FFNN) for classification, and a real mean square (rms) voltage,
peak voltage, crest factor, and total harmonic distortion (THD) to quantify the disturbances.
The techniques have been integrated into a methodology that allows the online processing
of a single PQD, and this has been validated and tested with synthesized signals and under
real operating conditions.
There are research papers that report propagation studies of specific disturbances;
in [11], the authors studied through simulations the propagation of only flicker phenomena
in a system supplied by a wind farm, and it was observed that, even if the wind turbines
are considered synchronized and producing the maximum level of voltage fluctuation,
the flicker indexes are below the limits imposed by current standards. The authors in [12]
realized a simulation to study the PQ and stability during a fault event propagated in a
microgrid using MATLAB. The outcome was that the placement of the fault and the type
greatly affect the stability of an autonomous microgrid. The fault propagation study was
performed only in simulation using MATLAB, so it has not been implemented in hardware.
In [13], the authors developed an efficient compressive sensing harmonics detector (CSHD)
to identify and estimate the principal pollution source of harmonics, which is simulated
and validated by means of appropriate testing to be performed on an example IEEE 13
bus distribution grid, thus obtaining an efficient and accurate CSHD for the identification
and estimation of the main harmonic sources in a grid. In [14], the authors analyzed
the impact of a distributed generation unit on a power grid, and then energy storage
systems (ESS) of different capacities were integrated into the power grid in an effort to
Sensors 2021, 21, 3910 3 of 21

study the improvements in the PQ. The obtained results showed that the integration of
energy storage systems into the power grid improves the PQ of the grid, which can be
extremely useful for system operators. In [15], the authors realized an economic study
to analyze the economic feasibility for the integration of flywheel ESS in a wind power
plant. The integration of ESS in the wind farm will allow an increase in the load factor of
the power plant by cutting down the probability of being disconnected from the power
grid for impacting the stability of the network. However, the integration is only feasible
with the government subsidy in renewable energy projects. The above-mentioned works
perform propagation and source detection studies of PQDs in simulations, so it is necessary
to implement some disturbance detection techniques in hardware. The authors in [16]
used tens of thousands of balancing electricity meters for measuring the quality of the
electricity supply indicator and identifying the source of PQDs based on the analysis of
10 min added data from the distributed measurement system. This article shows that
the system of the balance meter equipped with the PQ functionalities can provide a wide
spectrum of grid monitoring and diagnosis capabilities. However, this study is not always
suitable for other cases as it uses high-cost commercial equipment. The authors in [17]
presented a feasibility study on the procedure to implement PQ metrics in a low-cost smart
metering platform by the use of commercial modules for PQ measurement, and performed
a harmonic analysis. The results were delivered by choosing a real case study of STCOMET,
a commercial electronic board widely used for remote energy metering purposes, and
the paper has verified the possibility to implement PQ metrics in it. The authors of [18]
designed an intelligent street lighting PQ monitoring system to test the adaptive current
control strategy, in which a commercial measurement system from National Instruments
(NI) cDAQ-9185 was used. As a result, they observed that the THD decreased by 19%
on average. The above-mentioned works aimed to identify the main sources of a specific
disturbance in each case by the use, in some occasions, of high-cost commercial equipment.
However, it is important to monitor all disturbances and study how they propagate in the
grid, since this is essential to obtain an accurate synchronization between the devices that
allows the monitoring of different points of the grid at the same time.
There are initiatives to monitor disturbances in the grid, but they present some disad-
vantages. This notion PV-on time represents a local definition, in [19], and in fact is about
real-time monitoring of the plant, which has been developed to supervise the operation
mode of a Grid-Connected Utility-Scale photovoltaic power plant in order to ensure the
reliability and continuity of its supply. This system utilizes sensors distributed in the plant
to obtain a PQ analysis in real-time; the data acquisition equipment has been integrated
with a precision time protocol (PTP) to synchronize the collected data with a nanosecond
level of synchronization, yet it has the disadvantage that the integration turns out to be
an expensive wired infrastructure. The authors in [20] developed an open architecture
smart sensor network, based on field-programmable gate array (FPGA) technology, which
is capable of monitoring PQ continuously in industrial facilities, public buildings, and resi-
dential buildings. It is also capable of estimating different PQ indices, as well as identifying
disturbances and detecting connection and disconnection events, and it is also capable
of locating events in a synchronized way in different points of an electrical installation
by using a real-time clock (RTC) that allows a synchronization of the measurements in
different points of the grid. Nevertheless, this synchronization has problems because
the RTC of each device has different oscillators which, during long time measurements,
cause a desynchronization. The authors of [21] developed a measurement technique to
detect rapid voltage changes (RCV) and their propagation effect in an electrical distribution
grid; this technique has the ability to detect RCV disturbances and correlate them with
other electromagnetic interference events simultaneously, where four PQ analyzers, model
PQube 3 manufactured by PSL—Power Standards Lab, were used for data acquisition
while Raspberry Pi-3 devices were used to obtain the time reference, obtaining as a result
the probability that an RCV in an electrical distribution grid generates a sag in an uninter-
rupted source—70% to 93.8%. Using the Raspberry Pi as a time reference involves using the
Sensors 2021, 21, 3910 4 of 21

ethernet or WIFI port to keep the Raspberry Pi clock synchronized, which means additional
cost to the grid infrastructure. The authors in [22] used a phasor measurement unit (PMU)
for harmonic state estimation for an unbalanced three-phase distribution system. A PMU
utilizes a global positioning system (GPS) to perform synchronized measurements. This
allows the identification of harmonic sources and the monitoring of harmonic components,
and their propagation in the grid. PMUs are devices that estimate the magnitude and
phase angle of the voltage and current signals in the power grid, and are high-cost devices.
Owing to the limitations of the devices, it is important to develop a system capable of
synchronizing measurements that can accurately detect and classify PQDs, and that can
detect whether a disturbance has propagated throughout the network.
The aim of this work is to develop a system to detect whether a disturbance is gener-
ated at a point in the grid, and to detect to which point it has propagated within the grid in
the industrial facility; a proprietary system based on an FPGA sensor has been developed
to perform synchronized measurements with other proprietary systems using a GPS to
track the propagation of PQDs. The proposed method uses a parameterized model based
on genetic algorithms (GAs) and particle swarm optimization (PSO), which describes the
components of all PQDs to decompose the current or voltage signal to detect disturbances.
GPS allows the synchronization of the measurements acquired by the data logger, called
GSD in the experimental setup. This synchronization allows the tracking of the propa-
gation of disturbances within the grid. This method has been validated in an industrial
facility, installing three systems at different points of the network to detect disturbances
propagating through it, and an analysis of the propagated disturbances (transients, voltage
fluctuation, and harmonic content) has been carried out.
The contribution of this work is summarized in a punctual form as shown below.
• This paper describes a low-cost proprietary system which implements a methodology
to track the propagation of PQDs by performing synchronized measurements, using
GPS, between different proprietary data loggers located at different points in the
grid. This is due to the fact that other systems perform punctual PQ analysis of a
certain number of PQDs, and some other works do not use GPS, using other synchro-
nization techniques which have limitations that compromise the synchronization of
the measurements.
• This paper shows the hardware implementation of a full PQD parameterized model
based on PSO and GAs that allows the detection and classification of several PQDs.
Other works only allow the detection of a limited number of PQDs and perform
offline detection.
The contents of this paper are structured in different sections: Section 2 presents the
theoretical background, and some concepts that are an important part of the theory within;
Section 3 presents the hardware used in the paper; Section 4 presents the development of
the methodology for the detection and tracking of disturbances; Section 5 shows the results
obtained from the measurements performed in the industrial facility; Section 6 discusses
the results obtained; Section 7 presents the conclusions of this paper.

2. Theoretical Background
2.1. PQD
PQ is a concept which refers to several parameters such as voltage, current, and
frequency that are defined in a range. Any alteration of these parameters is known as
a PQD [6]. There are different types of disturbances, such as voltage sag, voltage swell,
transient, flicker, harmonics, interruption, among others [23].
Sensors 2021, 21, 3910 5 of 21

Each PQD is described by means of a mathematical model. However, the authors

in [7] have developed a mathematical model called a “full PQD parameterized model” that
describes all disturbances in one equation shown below.

x (t) = XCD + A[1 + δ(t)]{cos[2π f 0 (t)t + θ1 (t)]

N
+ ∑ ah (t) cos[2πh f 0 (t)t + θh (t)]}
h =2
k
+ ∑ bk (t) cos[2π f k (t)t + ϕk (t)] (1)
k =1
M h i
+ ∑ cm [u(t − αm ) − u(t − β m )] ∗ e − t−τmαm
m =1
∗ cos[2π f m t + ψm ] + n(t, x0 , σ) + µ(t)

Equation (1) classifies PQDs into five types of phenomena defined by their nature;
phenomena related to the fundamental frequency amplitude and harmonics, phenomena
related to the fundamental frequency with changes in frequency and phase, stationary
phenomena non-correlated to the fundamental frequency, transient phenomena, and addi-
tive random noise, Gaussian and non-Gaussian [7]. The first term of Equation (1), XCD ,
corresponds to the DC component. The second term has a parameter that controls the
behavior of the amplitude A[1 + δ(t)], in a fundamental component with a frequency f 0 (t),
amplitude A, and time-variant phase. θ1 (t). The third term corresponds to correlated har-
monic distortion, which includes time-variant amplitude ah (t) and phase θh (t). The fourth
term describes the non-correlated frequency components, which have time-variant ampli-
tudes bk (t), phase ϕk (t), and frequencies f k (t). The fifth term is related to fast transients,
i.e., short-time transients, between twice as many marks αm , and β m . These transients
include coefficients with amplitude cm and have an exponential or an oscillatory decaying
behavior τm , and the decay has a rate defined by a single frequency f m . The sixth term
describes Gaussian noise defined in two parameters: x0 is the statistical means, and σ is the
standard deviation. Finally, the seventh term models non-Gaussian noise as a time-variant
function [1].
Equation (1) has the advantage of modeling many types of disturbances, including
sags, swells, transients, harmonics, etc. For this work, transient impulsive, voltage fluctua-
tion, and harmonic content disturbances have been found in measurements performed in
industrial facilities, and are used to validate the system proposed in this work.

2.2. Transient
According to the IEEE std 1159 definition in [24], a transient is a phenomenon which
varies between two consecutive steady states during a short time interval; this disturbance
is classified into two categories: impulsive transient and oscillatory transient. In Figure 1,
an impulsive transient is shown, being a sudden, non-power frequency change of the
nominal voltage or current condition; this disturbance is unidirectional in polarity, and is
normally characterized by a short time of rise and a decay. Lightning is the most common
cause of impulsive transients. On the other hand, in Figure 2 an oscillatory transient is
shown; this is a sudden, nonpower frequency change in the steady-state condition of
voltage or current, which includes polarity changes [23]. The transient oscillation occurs
in a high-frequency area, and has a characteristic given by a short duration and a wide
range of transient frequency domain. Transient faults can lead to the breakdown of a line
insulation, power outages, and other serious problems [25].
Sensors
Sensors 2021,
Sensors2021,
Sensors 2021, 21,
2021,21,
21, xxxFOR
FOR
21,3910 PEER
FORPEER REVIEW
PEERREVIEW
REVIEW 666of 21
of21
of 21
21

Figure
Figure 1.
Figure1. Current
1.Current signal
Currentsignal which
signalwhich contains
whichcontains an
containsan impulsive
animpulsive transient.
transient.
impulsivetransient.
transient.
Figure 1. Current signal which contains an impulsive

Figure
Figure 2.
Figure2.
2.AA current
Acurrent signal
currentsignal which
signalwhich contains
whichcontains an
containsan oscillatory
anoscillatory transient.
transient.
oscillatorytransient.
transient.
Figure 2. A current signal which contains an oscillatory

2.3. Voltage
Voltage Fluctuation
2.3.Voltage Fluctuation
According
According
Accordingto toIEEE
IEEEstd
std1159 definition
1159definition
definitionin in
in[24], voltage
[24],voltage fluctuation
voltagefluctuation
fluctuationis is
isaaaseries
series
seriesofof voltage
ofvoltage
voltage
changes in the voltage
changes or a cyclical variation in the voltage envelope, as shown in Figure 3, onlamps
or a cyclical variation voltage envelope, as shown in Figure
Figure 3,
3, on
on lamps
lamps
such
such that
such that they
that they are
they are perceived
perceived
are perceived to
to
perceived to flicker
flicker
to flicker by
by
flicker by the
the
by the human
human
the human eye,
eye,
human eye, causing
causing
eye, causing irritation
irritation
causing irritation and
and
irritation and medical
medical
and medical
medical
problems
problems [26,27].
problems [26,27]. Sometimes,
[26,27]. Sometimes, voltage
Sometimes, voltage fluctuations
fluctuations
voltage fluctuations
fluctuations areare caused
are caused
caused byby an
by an arc
an arc furnace,
arc furnace, motors,
furnace, motors,
motors,
rolling
rolling mills,
mills, mash
mash welders,
welders, and
and
and electric
electric
electric welders
welders
welders
rolling mills, mash welders, and electric welders [28]. [28].
[28].
[28].

Figure
Figure 3.
Figure3. Signal
3.Signal of
Signalof voltage
ofvoltage fluctuation.
voltagefluctuation.
fluctuation.
Sensors 2021,21,
Sensors2021, 21,3910
x FOR PEER REVIEW 77of
of21
21

2.4. Harmonic
2.4. Harmonic Content
Content
In regard
In regard to
tothe
theIEEE
IEEEstdstd1159
1159definition
definitioninin[24],
[24],harmonic
harmonic content,
content, shown
shown in in Figure
Figure 4,4,
relies on sinusoidal
relies sinusoidal voltage
voltageororcurrent
currentfrequencies
frequencieswhich
whichareare
integer multiples
integer of the
multiples of fun-
the
damental frequency
fundamental frequencyof of
thethe
grid.
grid.The
Thefocal
focalsources
sourcesof ofharmonics
harmonics are devices,
are switching devices,
saturable
saturable devices,
devices, and
and arcing
arcing devices
devices [29].
[29]. The
The presence
presence ofofharmonics
harmonics in inthe
thegrid
gridcauses
causes
malfunction, increasesreliability
malfunction, and increases reliabilityproblems
problemsofof thethe power
power systems.
systems. Sensitive
Sensitive loadsloads
con-
connected to the grid are greatly affected by voltage imbalances and harmonics.
nected to the grid are greatly affected by voltage imbalances and harmonics. In the pres- In the
presence of harmonics,
ence of harmonics, motors
motors suffer
suffer fromfrom overheating,
overheating, reducing
reducing their their lifetime,
lifetime, and inand in
trans-
transmission
mission lineslines
they they
causecause
powerpower imbalance
imbalance and heating
and heating oflines
of the the lines
[30].[30].

Figure4.
Figure 4. Current
Currentsignal
signalwith
withharmonic
harmoniccontent.
content.

2.5.
2.5. Wavelet
Wavelet Transform
Transform(WT)
(WT)
AA wavelet is a small wave
wavelet is a small wave with
with oscillatory
oscillatory conditions
conditions from
from which
which thethe wavelet
wavelet func-
func-
tions
tions are generated by the use of a prototype mother wavelet, and a translation ττ that
are generated by the use of a prototype mother wavelet, and a translation that
corresponds
correspondstotoa awindow
window shifting.
shifting.The WTWT
The is depicted in Equation
is depicted (2), which
in Equation is a time-scale
(2), which is a time-
decomposition technique,
scale decomposition also used
technique, alsotoused
obtain tothe parameter
obtain of an input
the parameter signal
of an at different
input signal at
frequencies; they separate it into their different components [31].
different frequencies; they separate it into their different components [31].
1 t − nb0 am
(t), =( )p= m ( ( m 0 ))
Ψm,nΨ (2)
(2)
a0 a0

where m
where m and
and nn control
control the
the dilation
dilation and
and translation
translation of
of the
the wavelet,
wavelet, a0 isisaadilation
dilation step
step
parameter, is the location parameter, and Ψ ( ) is the mother wavelet.
parameter, b0 is the location parameter, and Ψm,n (,t) is the mother wavelet.

2.6. Hilbert–Huang
2.6. Hilbert–Huang Transform
Transform(HHT)
(HHT)
HHT
HHTisisaa tool
tool used
used to
to track
track and
and envelope
envelope signals,
signals,being
beingable
abletotodescribe
describethe
theamplitude
amplitude
changes
changesof
ofaagiven
givenwaveform.
waveform. Equations
Equations (3)–(5)
(3)–(5)describe
describethis
thistechnique.
technique.

z(t)(=) x=(t)(+) jx
+HT (t)( =
)=Aenv (t)( exp
)exp ( (t())))
( jθ (3)
(3)
( )q = ( ) + ( ) (4)
Aenv (t) = x (t)2 + x HT (t)2 (4)
( )
( )= ( ) (5)
x HT ((t ))
θenv (t) = tan−1 ( ) (5)
x (t)
The practical application of the HHT takes place throughout the analytical signal
( ) The practical
presented in application of the
Equation (3), HHT takes
where ( ) place
is thethroughout the analytical
envelop signal of ( ) as signal z(tin
stated )
presented in Equation
Equation (4), and (3),
( ) where Aenv (t) is the envelop
is the instantaneous phase ofsignal asxis(tshown
( ) of ) as stated in Equation
in Equation (5).
(4), and θenv (t) is the instantaneous phase of x (t) as is shown in Equation (5).
2.7. Genetic Algorithms (GAs)
Sensors 2021, 21, 3910 8 of 21

2.7. Genetic Algorithms (GAs)

GAs are among the most evolved and efficient classes of evolution inspired methods
and are based on the key principles of natural evolution theory, and they are also a powerful
meta-heuristic technique mainly used in problems where specific values are required, and
in which the conditions of the problem are complex. They are also applied where there is
lack of data or when previous knowledge on a specific problem is unknown. In such a way,
in which parameterized models or functions are impossible to define, another application
is the problem. The characteristics of that issue, such as a wide design space, nonlinearity,
non-convexity, and multi-objectivity, make it impossible for classical methodologies to find
an adequate solution [32].

2.8. Particle Swarm Optimization (PSO)

PSO is a technique referred to as a population-based stochastic optimization technique,
also used for online and offline monitoring to extract the best subset of features using an
extreme learning machine. This technique, in combination with the WT and the HHT,
allows finding those parameters of the proposed model, whereas GAs do not.

3. Proprietary PQD Detection System

To monitor the propagation of PQDs, it is necessary to develop a system that can
detect disturbances and perform synchronized measurements with other devices. The
proprietary data logger used is based on Xilinx® FPGA Spartan6 (XC6SLX16, Xilinx, San
Jose, CA, USA) with a 16-bit analog-to-digital converter (ADC) with eight channels from
Texas Instruments® (ADS130E08) (Texas Instruments, Dallas TX, USA), delivering up to
8000 samples per second, per channel, and synchronizing the measured electrical signals
using a pulse-per-second (PPS) provided by a GPS module receiver (U-Blox, Thalwil canton
of Zürich, Switzerland). PPS is a broadcast time base synchronized to the satellite atomic
clock used to synchronize all data loggers’ internal time base; it obtains a synchronization
of the internal time bases of the data loggers. The proprietary data logger with GPS
synchronization performs the measurement of four voltage signals (three phases; called Va,
Vb, and Vc, and one neutral; called Vn) and four current signals (three phases; called Ia, Ib,
and Ic, and one neutral; called In); the signals obtained are stored in a micro-SD memory. It
supports a maximum micro-SD memory capacity of 120 GB. The proprietary data logger
has a sampling rate of 8000 samples per second and per channel. Figure 5 shows an image
of the hardware that constitutes the proprietary data logger developed; the GPS antenna is
used to catch the signal from the satellites, the GPS receiver module allows decoding and
interpreting the signal received by the GPS antenna, and the Bluetooth module (Qualcomm,
San Diego, CA, USA) allows the connection of the data logger with the interface developed
for mobiles, as shown in Figure 6, which allows starting and terminating the measurement,
as well as seeing in real time the measured voltage and current signals. The data acquisition
system allows the digitalization of the eight channels of the input signals, the FPGA board
with the algorithms also allows the synchronization of the data logger and the detection
of disturbances (also shown), and the primary current sensors allow the detection of the
current through a conductor in a non-invasive way, but other current and voltage sensors
can be used with a voltage output signal.
Sensors 2021, 21, 3910 9 of 21
Sensors 2021, 21, x FOR PEER REVIEW 9 of 21

Sensors 2021, 21, x FOR PEER REVIEW 10 of 21

detection. It allows the determination of whether a disturbance has been detected simul-
taneously in different points within the grid. Figure 6b shows the real time display screen
of the signals being measured, allowing the user to select to view one input channel or
Figure5.5.
Figure
several inputImage
Image ofthe
of theproprietary
channelsproprietary GPS(GSD).
GPS (GSD).
simultaneously, distinguishing each signal with a different color.

The proprietary data logger hardware has been selected based on the purpose and
economy of the device—it is important to maintain the cost–benefit ratio without compro-
mising quality. To show the purpose of each specific component used, Table 1 shows the
model of each component, its main features, and its functionality for the proprietary data
logger.

Table 1. Proprietary data logger hardware technical specifications.

Component Model Feature Functionality

14579 Logic Elements. It is the main core of the data logger—it coor-
Xilinx® FPGA Spartan6 186 Input/Output. dinates all the modules, controls the ADC ac-
FPGA
(XC6SLX16) Voltage supply: 1.2 V. quisition, synchronizes its internal base with
Integrate Memory: 576 kbit. the PPS, and processes the measured signals.
16-bit conversion. The ADC performs the conversion of electrical
8 input channels. signals, this ADC contains 8 channels which
Texas Instruments®
ADC 8000 samples per second. are used to measure three-phase electrical sig-
(ADS130E08)
Interface Serial Peripheral Interface (SPI). nals, and its cost–benefit ratio is acceptable for
Architecture Delta-Sigma. the application.
Position accuracy: ± 2 m.
Communication protocol: National Marine
Electronics Association (NMEA), Radio Tech- This low-cost GPS module is responsible for
GPS module nical Commission for Maritime Services providing the PPS for the synchronization of
Blox Neo 6M-0-001
receiver (RTCM). measurements, and it is a generic module
Sensibility del receptor: −161 dBm. which is easily accessible to the public.
Voltage supply: 2.7–3.6 VCC.
Provides PPS signal.
Bluetooth 2.4 GHz. The Bluetooth module maintains communica-
Bluetooth BlueCore 4-External Enhanced data rate to 3 Mbps. tion with the mobile interface, the cost–benefit
module BC417143BGO + 6 dBm Radio frequency transmit power. ratio is appropriate for the application, and it
Standard Human Computer Interaction (HCI). is easily accessible to the public.

Figure 6 depicts the screens of the interface developed for mobile devices. This ap-
plication has been developed in C language, under the linux environment, up to now en-
abled to be used only for devices with Android operating system. Figure 6a shows the
control screen, which is used to initialize and finalize the measurement and, in addition,
the initial configuration of gain and offset of each reading channel (Ia, Ib, Ic, In, Va, Vb,
Figure 6.
Figure Imageinterface
6. Image interface of Vc,application
ofthe
the Vn), and developed
application the data of
developed forthe
for measurement
mobiles
mobiles tocontrol
to status,
control the
the date, data
proprietary
proprietary time, and GPS
data logger.
logger. location are dis-
(a) Measurement
(a) Measurement
control
control screen and configurationplayed. The developed
of measurement interface
channels: also
Ia, Ib, Ic, In, provides
In, Va,
Va, Vb, Vc, information
Vn. (b) about
(b) Real-time
Real-time thedisplay
signal
signal detected
display disturb-
screen.
screen.
ance: it indicates the time and which GPS synchronized data logger (GSD) performed the
The procedure used to synchronize the measured electrical signals is shown in Figure 7,
which describes the procedure of the internal time base synchronization: the first step’s
purpose is to verify if the PPS has been received by the GPS receiver module; in the second
step, the PPS information is stored in a flag called SMP; in the third stage, the internal time
base of the data logger is synchronized using the PPS provided by the GPS receiver mod-
Sensors 2021, 21, 3910 10 of 21

The proprietary data logger hardware has been selected based on the purpose and
economy of the device—it is important to maintain the cost–benefit ratio without com-
promising quality. To show the purpose of each specific component used, Table 1 shows
the model of each component, its main features, and its functionality for the proprietary
data logger.

Table 1. Proprietary data logger hardware technical specifications.

Component Model Feature Functionality

It is the main core of the data
14579 Logic Elements. logger—it coordinates all the
Xilinx® FPGA Spartan6 186 Input/Output. modules, controls the ADC
FPGA
(XC6SLX16) Voltage supply: 1.2 V. acquisition, synchronizes its
Integrate Memory: 576 kbit. internal base with the PPS, and
processes the measured signals.
16-bit conversion. The ADC performs the conversion
8 input channels. of electrical signals, this ADC
Texas Instruments® 8000 samples per second. contains 8 channels which are used
ADC
(ADS130E08) Interface Serial Peripheral to measure three-phase electrical
Interface (SPI). signals, and its cost–benefit ratio is
Architecture Delta-Sigma. acceptable for the application.
Position accuracy: ±2 m.
Communication protocol: National
This low-cost GPS module is
Marine Electronics Association
responsible for providing the PPS
(NMEA), Radio Technical
for the synchronization of
GPS module receiver Blox Neo 6M-0-001 Commission for Maritime
measurements, and it is a generic
Services (RTCM).
module which is easily accessible to
Sensibility del receptor: −161 dBm.
the public.
Voltage supply: 2.7–3.6 VCC.
Provides PPS signal.
Bluetooth 2.4 GHz.
The Bluetooth module maintains
Enhanced data rate to 3 Mbps.
communication with the mobile
BlueCore 4-External + 6 dBm Radio frequency
Bluetooth module interface, the cost–benefit ratio is
BC417143BGO transmit power.
appropriate for the application, and
Standard Human Computer
it is easily accessible to the public.
Interaction (HCI).

Figure 6 depicts the screens of the interface developed for mobile devices. This
application has been developed in C language, under the linux environment, up to now
enabled to be used only for devices with Android operating system. Figure 6a shows the
control screen, which is used to initialize and finalize the measurement and, in addition, the
initial configuration of gain and offset of each reading channel (Ia, Ib, Ic, In, Va, Vb, Vc, Vn),
and the data of the measurement status, date, time, and GPS location are displayed. The
developed interface also provides information about the detected disturbance: it indicates
the time and which GPS synchronized data logger (GSD) performed the detection. It allows
the determination of whether a disturbance has been detected simultaneously in different
points within the grid. Figure 6b shows the real time display screen of the signals being
measured, allowing the user to select to view one input channel or several input channels
simultaneously, distinguishing each signal with a different color.
The procedure used to synchronize the measured electrical signals is shown in
Figure 7, which describes the procedure of the internal time base synchronization: the first
step’s purpose is to verify if the PPS has been received by the GPS receiver module; in
the second step, the PPS information is stored in a flag called SMP; in the third stage, the
internal time base of the data logger is synchronized using the PPS provided by the GPS
receiver module; in the fourth step, the FPGA processor acquires the eight input channels
utilizing the 8 CH data acquisition system through the SPI communication protocol; and in
Sensors 2021, 21, 3910 11 of 21

the fifth step, the universal time coordinate (UTC) time is updated using the universal asyn-
Sensors 2021, 21, x FOR PEER REVIEW
chronous receiver-transmitter (UART) communication protocol and stored in micro-SD11 of 21
Sensors 2021, 21, x FOR PEER REVIEW
memory, and the cycle is repeated. 11 of 21

Figure7.7.Proprietary
Proprietary datalogger
logger withGPS
GPSsynchronization
synchronizationdiagram.
diagram.
Figure 7. Proprietary data
Figure data logger with
with GPS synchronization diagram.

4.4.Methodology for
forPropagation Detection of PQDs
4. Methodology
Methodology for PropagationDetection
Propagation Detectionof ofPQDs
PQDs
The
The methodology for the monitoring monitoring of PQD PQDpropagation
propagation isperformedperformedby by theuse use
The methodology
methodology for the monitoring ofofPQD propagation isisperformed by thethe use of of
of
thethe GPS data logger synchronization procedure reported in Section 3, well
as well as the
the GPS data logger synchronization procedure reported in Section 3, as well as the fullfull
GPS data logger synchronization procedure reported in Section 3, as as the
full parameterized
parameterized PQD PQDmodel model reported
reported in [1,7],
in[1,7],
[1,7], in which a hybrid method based on GAs
parameterized PQD model reported in ininwhich
which a ahybrid
hybrid method
method based
based onon GAsGAs andand
and
PSO PSO
for for
the the estimation
estimation of the of the parameters
parameters of of
Equation Equation
(1) is (1)
used is
to used
automate to automate
the the
detection
PSO for the estimation of the parameters of Equation (1) is used to automate the detection
detection
process. process. The methodology PQDfor PQD propagation detection hasimplemented
been implemented
process. The methodology
The methodology forPQD
for propagation
propagation detection
detection hashasbeen
been implemented onon thethe
on
FPGAthe FPGA
containedcontained
in the in the
proprietaryproprietary
GPS GPS synchronized
synchronized data data
logger logger
(GSD), (GSD),
as shownas shown
in Fig-
FPGA contained in the proprietary GPS synchronized data logger (GSD), as shown in Fig-
in
ure Figure
8; the 8; the diagram
diagram shows shows
the the general
general blocks blocks
that that integrate
integrate the the proprietary
proprietary PQD PQD
detection
ure diagram shows the general blocks that integrate the proprietary PQD detection
detection
system. system.
system. The Bluetooth
The Bluetooth
Bluetooth moduleisis
module
module is responsible
responsible
responsible forreceiving
for
for receiving
receiving the instructions
theinstructions
the instructions from
from thethe
from
mo-mo-
the
bile mobile application,
bile application, and and
and transmitting transmitting
transmittingthe the
themeasured measured
measuredsignal signaldatasignal
datafor data for visualization.
forvisualization.
visualization. TheTheGPS GPS The
re-re-
GPS
ceiverreceiver
ceiver module module is responsible
responsible
is responsible for for providing
forproviding
providing thePPS
the thesynchronized
PPS PPS synchronized
synchronized totothethetodata
datathelogger.
data logger.
logger. TheThe
The current
current
current andand voltage
voltage
voltage inputinput
input signals
signals
signals are
are are detected
detected
detected by
by by the
the
the primarysensors
primary
primary sensorsfor
sensors for
for conditioning
conditioning
conditioning
and
and digitizing in
and digitizing in the the 8-channel
the 8-channel
8-channeldata data acquisition
dataacquisition system.
acquisitionsystem.
system.The The digitized
Thedigitized
digitized current
current
current and
andand voltage
voltage
voltage
signals
signals
signals are
are processed
processed by the signal feature extraction block, which is implementedinside
processed by
by the
the signal
signal feature
feature extraction
extraction block,
block, which
which is
is implemented
implemented inside
inside
the
theFPGA processor, to estimate the parameters of of Equation(1), (1), andthe thePQDPQDdetection
detection
the FPGA processor,
processor, to to estimate
estimatethe theparameters
parameters ofEquationEquation (1),and and the PQD detection
block
blockisisresponsible
responsible for for disturbance detectiondetection using the the parametersextracted extractedfrom fromthe the
block responsible for disturbance
disturbance detectionusing using theparameters
parameters extracted from the
previous block, obtaining
previous block, obtaining a flag that indicates whether a disturbance has been detected and
previous obtaining aa flag flag that
thatindicates
indicateswhether
whethera adisturbance
disturbancehas has been
been detected
detected
the
andsample number
the sample
sample of the location
number of the disturbance in the signal.
and the number of of the
thelocation
locationofofthe thedisturbance
disturbanceininthe thesignal.
signal.

Figure
Figure 8. 8. Diagram
Diagram of of disturbance
disturbance propagation
propagation monitoring
monitoring procedure
procedure using
using GPS-synchronized
GPS-synchronized data
data loggers
loggers called
called GSDGSD
and
Figure
and 8. parameterized
full Diagram of disturbance
PQD propagation monitoring procedure using GPS-synchronized data loggers called GSD
model.
full parameterized PQD model.
and full parameterized PQD model.
Sensors
Sensors2021,
2021,21,
21,x3910
FOR PEER REVIEW 12
12of
of21
21

Figure
Figure99shows
showsthe theblock
blockdiagram
diagramof ofthe
thesignal
signalfeature
featureextraction
extractionand
andPQD
PQDdetection
detection
sections
sections in general. The diagram has a voltage or current input signal: in thefirst
in general. The diagram has a voltage or current input signal: in the firstblock,
block,
the
the signal
signal isis segmented
segmented into into different
different frequencies,
frequencies, using
using techniques
techniquessuch
such as
as low-pass,
low-pass,
band-pass,
band-pass,and andhigh-pass
high-passfilters. Then,
filters. Then,thethe
signal provided
signal by the
provided by band-pass filter filter
the band-pass is seg-is
mented
segmented by the WT,WT,
by the using
usingEquation
Equation(2),(2),
to to
separate
separate the transients
the transientsfrom
fromthe
theharmonic
harmonicand and
interharmonic content, which are separated using the fast Fourier
interharmonic content, which are separated using the fast Fourier transform (FFT). Thetransform (FFT). The
second
secondblock
blockof ofthe
thediagram
diagramisis responsible
responsible for for classifying
classifying the
the disturbances
disturbances according
according to to
the
the signals segmented in the previous block, and Equations (3)–(5) of the HHTare
signals segmented in the previous block, and Equations (3)–(5) of the HHT are used
used
toto detect
detect ifif there
there are
are oscillations,
oscillations, voltage
voltage fluctuations,
fluctuations, and and signal
signal interruptions
interruptions coming
coming
from
fromthethelow-pass
low-passfilter. GAs
filter. GAsareare
used to classify
used the harmonic
to classify the harmoniccontent and transients,
content PSO
and transients,
isPSO
usedis to classify
used the interharmonic
to classify content,
the interharmonic and theand
content, statistical analysis
the statistical allows allows
analysis the classi-
the
fication of the of
classification additive Gaussian
the additive noise. noise.
Gaussian

Blockdiagram
Figure9.9.Block
Figure diagramof
ofextraction
extractionand
anddetection
detectionof
ofPQDs.
PQDs.

Figure 10
Figure 10 shows
shows thethe corresponding
correspondingsteps stepstotodetect
detectthethe
propagation
propagation of PQDs
of PQDsin the
in grid
the
using the GSD, the first step being the installation of the GPS synchronized
grid using the GSD, the first step being the installation of the GPS synchronized data log- data logger in
different
ger locations
in different of interest
locations in the grid.
of interest in theThe second
grid. The step
secondconsists of the initialization
step consists of the
of the initializa-
tion of the monitoring of each GPS synchronized data logger, and continues with the thirdis
monitoring of each GPS synchronized data logger, and continues with the third step that
the synchronization
step of the internal
that is the synchronization time
of the base with
internal timethe PPS
base of each
with dataof
the PPS logger; at this
each data stage,
logger;
the signals measured are synchronized. The fourth step is based on the
at this stage, the signals measured are synchronized. The fourth step is based on the meas- measurement and
storage of the eight input channels of the synchronized GPS. The fifth
urement and storage of the eight input channels of the synchronized GPS. The fifth step step consists of the
analysis of the voltage and current signals using the full PQD parameterized model; in
consists of the analysis of the voltage and current signals using the full PQD parameter-
this step, the hybrid approach processes the voltage and current signals, and performs its
ized model; in this step, the hybrid approach processes the voltage and current signals,
decomposition using WT, FFT, HHT, GAs, PSO, and statistical analysis, the purpose of
and performs its decomposition using WT, FFT, HHT, GAs, PSO, and statistical analysis,
which is to analyze it term by term. The model parameters are estimated by minimizing
the purpose of which is to analyze it term by term. The model parameters are estimated
the error between the original signal under analysis and the parameterized signal from
by minimizing the error between the original signal under analysis and the parameterized
the analytical of Equation (1). In the sixth step, the parameters obtained in the previous
signal from the analytical of Equation (1). In the sixth step, the parameters obtained in the
step are used to determine whether there has been a disturbance, and thus generate a flag
previous step are used to determine whether there has been a disturbance, and thus gen-
indicating the sample number of the disturbance. Finally, in the seventh step, the sample
erate a flag indicating the sample number of the disturbance. Finally, in the seventh step,
number and time of the disturbance found are used to determine the presence of the same
the sample number and time of the disturbance found are used to determine the presence
disturbance in the other data loggers, sending information obtained by the GSD to the
of the same disturbance in the other data loggers, sending information obtained by the
developed application, which compares the times amongst similar disturbances detected:
GSD
shouldto the developed
these times be application,
within a periodwhichof 5compares the times as
s, it is considered amongst similar disturb-
the propagation of that
ances
disturbance, and so, the internal time base is synchronized with the PPS as
detected: should these times be within a period of 5 s, it is considered to the propa-
repeat the
gation of that disturbance, and so, the internal time base is synchronized with the PPS to
cycle again.
repeat the cycle again.
Sensors 2021, 21, 3910 13 of 21
Sensors 2021, 21, x FOR PEER REVIEW 13 of 21

Figure10.
Figure 10. Diagram
Diagram of
of the
theprocedure
procedurefor
formonitoring
monitoringa adisturbance that
disturbance propagates
that in in
propagates thethe
grid im-
grid
plemented in GSD.
implemented in GSD.

5.5.Results
Results
Thissection
This sectionshows
showsthe
theresult
resultanalysis
analysisofofthe
theproposed
proposedmethodology
methodologyby byperforming
performing
thetest
the testwith
withdifferent
differentdisturbances
disturbancesdetected
detectedand
andtracked
trackedin inthe
thegrid
gridusing
usingthe
theprocedure
procedure
described
describedininthe
theprevious
previous section; thethe
section; first subsection
first describes
subsection the experimental
describes setup,
the experimental the
setup,
second subsection
the second describes
subsection the transient
describes analysis,
the transient the third
analysis, subsection
the describesdescribes
third subsection the voltage
the
fluctuation analysis, and the fourth subsection describes the harmonic content
voltage fluctuation analysis, and the fourth subsection describes the harmonic content analysis.
analysis.
5.1. Validation Setup
For the experimental
5.1. Validation Setup setup of this work, three proprietary GPS synchronized data
loggers,
For the experimental setupand
called GSD-1, GSD-2, GSD-3,
of this were
work, installed
three at different
proprietary locations in data
GPS synchronized the
industrial facility grid, as shown in the schematic diagram in Figure 11.
loggers, called GSD-1, GSD-2, and GSD-3, were installed at different locations in the in-
dustrial facility grid, as shown in the schematic diagram in Figure 11.
 Sensors 2021, 21, 3910 14 of 21
Sensors 2021, 21, x FOR PEER REVIEW 14 of 21

Figure 11.11.
Figure Diagram
Diagram of of
thethe
installation of of
installation thethe
GPS synchronized
GPS data
synchronized logger
data (GSD-1,
logger GSD-2,
(GSD-1, and
GSD-2, and
GSD-3) in different locations of the grid.
GSD-3) in different locations of the grid.

5.2.
5.2. Transient
Transient Analysis
Analysis
Figure
Figure 12a–c
12a–c show
show thethe propagation
propagation tracking
tracking ofof
anan impulsive
impulsive transient
transient disturbance;
disturbance;
this
this disturbance
disturbance is propagated
is propagated to twoto two locations
locations in theingrid
theshown
grid shown
in Figure in 11.
Figure 11. The
The impul-
impulsive
sive transienttransient
is causedisby caused by anmotor
an electric electric motoratstarting
starting the locationat the location
where GSD-1wherehas GSD-1
been
has been installed. This disturbance is propagated to the location
installed. This disturbance is propagated to the location of GSD-2, being the main distri- of GSD-2, being the
main distribution panel of the line where the electric motor in
bution panel of the line where the electric motor in question is located. Yet, this disturb- question is located. Yet,
thishas
ance disturbance has nottopropagated
not propagated the locationtoofthe thelocation of the GSD-3
GSD-3 because because
it is located onitanother
is locatedlineon
ofanother
the grid.line of the grid.
Figure
Figure 12a,b
12a,b show
show thethe characteristics
characteristics ofofthe
thepropagated
propagateddisturbance.
disturbance.The Theimpulsive
impulsive
transient observed in Figure 12a has an amplitude (δa) of
transient observed in Figure 12a has an amplitude (δa) of 19.05 A and a duration (δt) 19.05 A and a duration (δt)ofof
0.06 s. On the other hand, the disturbance in Figure 12b has an
0.06 s. On the other hand, the disturbance in Figure 12b has an amplitude (δa) of 19.63 amplitude (δa) of 19.63 A and
A
a duration (δt) of 0.06 s, which shows that the amplitude of the
and a duration (δt) of 0.06 s, which shows that the amplitude of the disturbance in Figure disturbance in Figure 12b
has an increase of 0.58 A compared with the disturbance
12b has an increase of 0.58 A compared with the disturbance in Figure 12a. However, thein Figure 12a. However, the
durationtimes
duration timesofofboth
bothtransients
transients remain
remain the the same.
same.TheThedisturbance
disturbanceininFigureFigure12b is delayed
12b is de-
by 0.15 s after it is caused. The start time (ti) of the disturbance in
layed by 0.15 s after it is caused. The start time (ti) of the disturbance in (a) takes place(a) takes place at 462.89
at s,
and the beginning time (ti) of the disturbance in (b) is at 463.04
462.89 s, and the beginning time (ti) of the disturbance in (b) is at 463.04 s; with ground ons; with ground on these
similar
these characteristics,
similar characteristics, it can be determined
it can be determined thatthat
it is it
theis same disturbance.
the same disturbance.
Sensors 2021, 21, x FOR PEER REVIEW 15 of 21

Sensors 2021, 21, 3910 15 of 21

Figure 12. Signals (a–c) correspond to GPS synchronized data logger GSD-1, GSD-2, and GSD-3, and
the graphics show an impulsive transient propagated to two locations on the grid.

Figure 12.The
Signals (a–c) correspond
detection to GPS synchronized
of this disturbance has beendata logger GSD-1,
performed usingGSD-2,
the fulland
PQDGSD-3, and
parameter-
the graphics show an impulsive transient propagated to two locations on the grid.
ized model of Equation (1), in which techniques have detected this disturbance using GAs
and PSO. The part of Equation (1) that represents this disturbance is shown in Equation (6),
The detection to
corresponding of the
thistransient
disturbance has been where
phenomena, performed using
αm , and β mthe full the
define PQD parameter-
time value when
izedthe
model of Equation
transient (1), in
starts and which
ends, techniquescm
respectively, have detected
is the amplitudethis disturbance
factor, f m is using GAs
the frequency
andvalue,
PSO. The part of Equation
and ψm is the phase value.(1) that represents this disturbance is shown in Equation
(6), corresponding to the transient phenomena, where , and define the time value
M
is thet −
 
when the transient starts and ends, respectively, amplitude
αm factor, is the fre-
quency value, and m∑
x (t) = cm [u(t − αm ) − u(t − β m )] ∗ e − ∗ cos[2π f m t + ψm ] (6)
=1
is the phase value. τ m

5.3. Voltage Fluctuation Analysis −

()= ( − )− − ∗ −
(6)this
Figure 13a–c display the propagation tracking of a voltage fluctuation disturbance;
=1
disturbance is propagated throughout the three locations in the grid shown in Figure 11.
∗ cos 2 +
Voltage fluctuation is caused by loads with high demand variation, such as inverters, arc
furnaces, etc. Accordingly, the voltage fluctuation is reflected in the location of GSD-1,
5.3. GSD-2,
Voltage and GSD-3.Analysis
Fluctuation
Figure 13a–c display the propagation tracking of a voltage fluctuation disturbance;
this disturbance is propagated throughout the three locations in the grid shown in Figure
 Figure 13a–c show the characteristics of the disturbance; in Figure 13a, the voltage
fluctuation has an amplitude (δa) of 0.7669 V and a period (T) of 120.56 s; for Figure 13b,
the voltage fluctuation has an amplitude (δa) of 0.6414 V and a period (T) of 120.56 s; and
for Figure 13c, the voltage fluctuation also has an amplitude (δa) of 0.6375 V and a period
Sensors 2021, 21, 3910 16 of 21
(T) of 120.56 s. As the disturbance characteristics at the three locations are similar, and
there is no delay between them, the disturbance is determined to be the same.

Figure
Figure13.
13.Signals
Signals(a–c)
(a–c)correspond
correspondto
toGPS
GPSsynchronized
synchronizeddata
datalogger
loggerGSD-1,
GSD-1,GSD-2,
GSD-2, and
andGSD-3,
GSD-3, and
and
show
showaavoltage
voltagefluctuation
fluctuation disturbance
disturbance in
in voltage
voltage that
that propagates
propagates to
to three
three locations.

For the detection of the voltage fluctuation, the methodology makes use of the part in
the full PQD parameterized model, from Equation (1), which corresponds to the phenomena
related to the amplitude of the fundamental frequency and its harmonics, represented by
Equation (7), where A is the peak value of the amplitude, δ(t) is a time dependent function
that stands for events associated with the amplitude disturbances such as oscillations,
voltage fluctuation, and interruptions, ah is the time-dependent amplitude factor, f 0 is
the value of the fundamental frequency, θ1 is the value of the phase for the fundamental
component, and h is the index value for the h-th harmonic.
" #
N
x1 (t) = A ∗ [1 + δ(t)] cos(2π f 0 t + θ1 ) + ∑ ah (t) cos(2πh f0 t + θh ) (7)
h =2

Figure 13a–c show the characteristics of the disturbance; in Figure 13a, the voltage
fluctuation has an amplitude (δa) of 0.7669 V and a period (T) of 120.56 s; for Figure 13b,
the voltage fluctuation has an amplitude (δa) of 0.6414 V and a period (T) of 120.56 s; and
for Figure 13c, the voltage fluctuation also has an amplitude (δa) of 0.6375 V and a period
(T) of 120.56 s. As the disturbance characteristics at the three locations are similar, and
there is no delay between them, the disturbance is determined to be the same.

5.4. Harmonic Content Analysis

Figure 14a–c show the propagation tracking of the harmonic content; this disturbance
is propagated to the three locations in the grid depicted in Figure 11. Harmonic content
 5.4. Harmonic Content Analysis
Figure 14a–c show the propagation tracking of the harmonic content; this disturbance
Sensors 2021, 21, 3910 17 of 21
is propagated to the three locations in the grid depicted in Figure 11. Harmonic content is
caused by non-linear loads which produce a signal distortion. In this regard, harmonic
content is propagated to the location of GSD-1, GSD-2, and to a lower proportion to GSD-3.
Theisplots in Figure
caused 14 illustrate
by non-linear loadsthe amplitude
which producespectrum
a signal corresponding to signals
distortion. In this regard, Figure
harmonic
14a–c, respectively.
content is propagated to the location of GSD-1, GSD-2, and to a lower proportion to
GSD-3. The plots in Figure 14 illustrate the amplitude spectrum corresponding to signals
Figure 14a–c, respectively.

Figure 14. Signals (a–c) correspond to GPS synchronized data logger GSD-1, GSD-2, and GSD-3, and
the graphics show a harmonic content tracking where it propagated to three sites in the grid.

Figure 14.Figure
Signals14a–c
(a–c) correspond to GPS synchronized
show the harmonic content in data logger
current GSD-1,
signals; inGSD-2,
Figureand GSD-3,
14a,b, the and
signals
the graphics
are veryshow a harmonic
similar, content
but Figure tracking
14c has where it propagated
a difference—this to threeaccording
is because, sites in theto
grid.
the location
of the data logger, GSD-1 and GSD-2 are on the same line, and GSD-3 is located on a
Figure 14a–c
different show the harmonic
line. However, through thecontent in current
analysis signals; in Figure
of the amplitude spectrum14a,b, the sig-
of each signal,
nalsitare very similar,
is observed but ones
that the Figure 14c has a difference—this
corresponding to 15a–b both is because,
have according
the same to theand
frequencies
location
similarof amplitudes.
the data logger, GSD-1 and
Nevertheless, theGSD-2 are onspectrum
amplitude the samecorresponding
line, and GSD-3 is located
to signal 15c has
on athe
different line. However, through the analysis of the amplitude spectrum
same frequencies but with a much smaller amplitude; this indicates that the of each sig-
harmonic
nal,content
it is observed
propagatesthat to
the
theones corresponding
three locations, but to 15a–b
in the bothlocation,
GSD-3 have thethe same frequencies
propagation occurs
andinsimilar amplitudes.
a smaller proportion. Nevertheless, the amplitude spectrum corresponding to signal
15c has the same frequencies
Equation but with
(7) also allows a much smaller
the detection amplitude;
of harmonics thismodel
as the indicates that the the
represents
phenomena related to the amplitude of the fundamental frequency and harmonics, which
is part of the full PQD parameterized model of Equation (1).
 harmonic content propagates to the three locations, but in the GSD-3 location, the propa-
gation occurs in a smaller proportion.
Equation (7) also allows the detection of harmonics as the model represents the phe-
Sensors 2021, 21, 3910 18 of 21
nomena related to the amplitude of the fundamental frequency and harmonics, which is
part of the full PQD parameterized model of Equation (1).

6. Discussion
6. Discussion
The resultsTheobtained
resultsshow
obtainedthe detection
show theof differentofPQDs
detection thatPQDs
different propagated to other to other
that propagated
locations inlocations
the grid.inFigure 12 shows
the grid. Figurea12
transient
shows athat is propagated
transient only to twoonly
that is propagated monitored
to two monitored
locations oflocations
the gridofbecause
the gridthebecause
transient
theistransient
due to the starting
is due of starting
to the an electric
of anmotor, andmotor, and
electric
this disturbance is generated
this disturbance is in a point of
generated inaabranch
point ofofathe grid of
branch and
thepropagated through it. through it.
grid and propagated
However, it did not propagate
However, to other branches.
it did not propagate to other branches.
Figure 13 shows
Figurea13 voltage
showsfluctuation
a voltage disturbance. In this, it canIn
fluctuation disturbance. bethis,
observed
it canthat the
be observed that
disturbance the
hasdisturbance has been
been propagated propagated
to the to the three
three monitored points monitored points
of the grid, which ofmeans
the grid, which
means
that this type of that this type can
disturbance of disturbance
propagate can propagate
between between
branches. branches.
This is caused Thisbyis the
caused by the
switching ofswitching
nonlinearof loads,
nonlinear
andloads,
can beand can be in
reflected reflected in the flickering
the flickering of the lighting.
of the lighting. Moreo- Moreover,
ver, FiguresFigures
14 and1415and 15 show
show the propagation
the propagation of harmonic
of harmonic content
content through
through all monitored points
all monitored
of grid.
points of the the grid.

Figure 15. Frequency

Figure 15.spectrum plots
Frequency of the plots
spectrum signals
of corresponding to (a–c) are shown.
the signals corresponding to (a–c) are shown.

The resultsThe resultscorroborate

obtained obtained corroborate that the developed
that the developed methodology methodology and the synchro-
and the synchro-
nization ofnization of measurements
measurements of electrical
of electrical signals allow signals allow of
the detection the detection
PQDs of PQDs
and their prop- and their
agation in apropagation in be
grid; this can a grid;
usedthis can be
to track theused to track the
propagation propagationtoofother
of disturbances disturbances
points to other
points within a grid.
within a grid.

7. Conclusions
The propagation of PQDs can cause severe damages to sensitive equipment connected
at long distances away from the disturbance’s point of origin. In this work, a system has
been developed that allows the detection of the propagation of electrical disturbances
Sensors 2021, 21, 3910 19 of 21

using a methodology based on a FPGA sensor that also allows measurements synchronized
with GPS.
The development of a proprietary system solves some limitations of other works,
shown in Table 2. They include the implementation of models and methodologies that
allow an online detection of several PQDs, in addition to methodologies that allow the
tracking of the propagation of disturbances with the synchronization of the measurements
using GPS.

Table 2. Performance comparison of the devices used in different works.

System Capabilities Disadvantages

PMU can directly measure frequency, voltage, and
current waveforms along with phase angle differences Does not capture the raw waveform of voltage
at high sampling rates, and with great accuracy. and current.
PMU [22]
PMU utilizes a GPS reference source to provide the Does not calculate PQD indices.
required synchronization across wide High cost per unit.
geographical areas.
Stores only PQDs detected, does not store the
PQ analyzer calculates some PQ indices in real time.
raw data of current and voltage signal.
PQ Analyzer [16,19,21] Records electrical signals when it detects
Performs punctual measurements in the grid.
a disturbance.
High cost per unit.
Continuous measurement of raw waveform of voltage
and current.
Manual recovery of raw data to
GPS synchronization with PPS obtaining 1 error
post-processing.
GSD, our proposal sample in 24 h during a PPS interruption.
Data storage limitation to 120 GB,
PQD detection on-line using GAs and PSO.
approximately 2 weeks of monitoring.
PQD propagation monitoring.
Low cost per unit.

The proposed system and methodology fully prove their performance by being tested
in an industrial installation, where disturbances such as transients, voltage fluctuation, and
harmonic content are detected and tracked; regarding transients, they propagated to one
site, whereas in the case of voltage fluctuation it propagated to two sites, and in the case
of harmonic content it propagated to two sites. For further work, we intend to use this
methodology to track PQDs in other areas or facilities where there are other types of loads
that generate other types of disturbances.
The integration and use of the PPS and the proprietary data logger allow the synchro-
nization of measurements taken at different sites in the grid, while the implementation of
disturbance detection techniques allows the detection of disturbances and the monitoring
of whether disturbances propagate to other sites in the grid. However, the system has the
limitation of memory for raw data storage to a micro-SD slot, which limits the disturbance
monitoring time up to about two weeks.
In further works, we intend to improve the interface of the application that allows
us to visualize and control the systems, and implement a second micro-SD memory to
extend the time of continuous monitoring. We also plan to perform more tests in industrial
facilities to detect other types of disturbances.

Author Contributions: O.N.P.-Z. developed the methodology presented, wrote most of the paper,
and generated the data; R.d.J.R.-T. conceived and developed the idea of this research and wrote
most of the paper; J.R.M.-A. contributed to the development of the methodology and wrote some
of the paper; D.M.-S. developed the experimental setup, generated the data, and wrote some of the
paper; R.A.O.-R. conceived and developed the idea of this research, developed the methodology
presented, and wrote most of the paper; J.A.A.-D. developed the experimental setup, generated the
data, and wrote some of the paper. All authors have read and agreed to the published version of the
manuscript.
Sensors 2021, 21, 3910 20 of 21

Funding: This research was funded by Conacyt (Consejo Nacional de Ciencia y Tecnologia), under
the scholarship 486847.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: No new data were created or analyzed in this study. Data sharing is
not applicable to this article.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Rodriguez-Guerrero, M.A.; Jaen-Cuellar, A.Y.; Carranza-Lopez-Padilla, R.D.; Osornio-Rios, R.A.; Herrera-Ruiz, G.; Romero-
Troncoso, R.D.J. Hybrid Approach Based on GA and PSO for Parameter Estimation of a Full Power Quality Disturbance
Parameterized Model. IEEE Trans. Ind. Inform. 2018, 14, 1016–1028. [CrossRef]
2. Wright, P.S.; Christensen, A.E.; Davis, P.N.; Larson, B.; Lippert, T.; Patel, P. Power quality propagation measurements in smart
grids. In Proceedings of the 2016 Conference on Precision Electromagnetic Measurements (CPEM 2016), Ottawa, ON, Canada,
10–15 July 2016; pp. 16–17. [CrossRef]
3. SEL-735 Power Quality Meter and Billing|Schweitzer Engineering Laboratories. Available online: https://selinc.com/es/
products/735/ (accessed on 8 May 2021).
4. Basic Three-Phase Power Quality Recorder Fluke 1760TR|Fluke. Available online: https://www.fluke.com/es-mx/producto/
comprobacion-electrica/calidad-electrica/1760tr-basic (accessed on 8 May 2021).
5. Chawda, G.S.; Shaik, A.G.; Shaik, M.; Padmanaban, S.; Holm-Nielsen, J.B.; Mahela, O.P.; Kaliannan, P. Comprehensive Review on
Detection and Classification of Power Quality Disturbances in Utility Grid with Renewable Energy Penetration. IEEE Access 2020,
8, 146807–146830. [CrossRef]
6. Mishra, M. Power quality disturbance detection and classification using signal processing and soft computing techniques: A
comprehensive review. Int. Trans. Electr. Energy Syst. 2019, 29. [CrossRef]
7. Rodriguez-Guerrero, M.A.; Carranza-Lopez-Padilla, R.; Osornio-Rios, R.A.; de J. Romero-Troncoso, R. A novel methodology for
modeling waveforms for power quality disturbance analysis. Electr. Power Syst. Res. 2017, 143, 14–24. [CrossRef]
8. Borrás-Talavera, M.D.; Bravo, J.C.; Álvarez-Arroyo, C. Instantaneous disturbance index for power distribution networks. Sensors
2021, 21, 1348. [CrossRef]
9. Divya, S.; Uma Rao, K. Comparative performance of different wavelets in Power Quality disturbance detection and quantification.
In Proceedings of the 2016 Second International Conference on Cognitive Computing and Information Processing (CCIP), Mysuru,
India, 12–13 August 2016; pp. 1–5. [CrossRef]
10. Granados-Lieberman, D.; Valtierra-Rodriguez, M.; Morales-Hernandez, L.A.; Romero-Troncoso, R.J.; Osornio-Rios, R.A. A Hilbert
transform-based smart sensor for detection, classification, and quantification of power quality disturbances. Sensors 2013, 13,
5507–5527. [CrossRef]
11. Lazaroiu, G.C.; Golovanov, N.; Balaban, G.; Corti, M.; Tironi, E.; Zaninelli, D. Flicker propagation in power systems supplying
wind farms. In Proceedings of the International Conference on Harmonics and Quality of Power, ICHQP, Belo Horizonte, Brazil,
16–19 October 2016; Volume 2016-Decem, pp. 66–70.
12. Pungut, N.A.F.B.; Hannoon, N.M.S.; Ibrahim, P. Bin Power Analysis of Autonomous Microgrid. J. Sci. Ind. Res. 2017, 76, 626–630.
13. Carta, D.; Muscas, C.; Pegoraro, P.A.; Sulis, S. Identification and Estimation of Harmonic Sources Based on Compressive Sensing.
IEEE Trans. Instrum. Meas. 2019, 68, 95–104. [CrossRef]
14. Nieto, A.; Vita, V.; Maris, T.I. Power Quality Improvement in Power Grids with the Integration of Energy Storage Systems. Int. J.
Eng. Res. Technol. 2016, 5, 438–443.
15. Nieto, A.; Vita, V.; Ekonomou, L.; Mastorakis, N.E. Economic Analysis of Energy Storage System Integration with a Grid
Connected Intermittent Power Plant, for Power Quality Purposes. WSEAS Trans. Power Syst. 2016, 11, 65–71.
16. Chmielowiec, K.; Wiczynski, G.; Rodziewicz, T.; Firlit, A.; Dutka, M.; Piatek, K. Location of power quality disturbances sources
using aggregated data from energy meters. In Proceedings of the 2020 12th International Conference and Exhibition on Electrical
Power Quality and Utilisation-(EPQU), Cracow, Poland, 14–15 September 2020; pp. 3–7. [CrossRef]
17. Artale, G.; Caravello, G.; Cataliotti, A.; Cosentino, V.; Di Cara, D.; Dipaola, N.; Guaiana, S.; Panzavecchia, N.; Sambataro, M.G.;
Tinè, G. Pq and harmonic assessment issues on low-cost smart metering platforms: A case study. Sensors 2020, 20, 6361. [CrossRef]
18. Martinek, R.; Bilik, P.; Baros, J.; Brablik, J.; Kahankova, R.; Jaros, R.; Danys, L.; Rzidky, J.; Wen, H. Design of a measuring system
for electricity quality monitoring within the smart street lighting test polygon: Pilot study on adaptive current control strategy for
three-phase shunt active power filters. Sensors 2020, 20, 1718. [CrossRef]
19. Moreno-Garcia, I.M.; Palacios-Garcia, E.J.; Pallares-Lopez, V.; Santiago, I.; Gonzalez-Redondo, M.J.; Varo-Martinez, M.; Real-Calvo,
R.J. Real-time monitoring system for a utility-scale photovoltaic power plant. Sensors 2016, 16, 770. [CrossRef]
20. Morales-Velazquez, L.; Romero-Troncoso, R.J.; Herrera-Ruiz, G.; Morinigo-Sotelo, D.; Osornio-Rios, R.A. Smart sensor network
for power quality monitoring in electrical installations. Meas. J. Int. Meas. Confed. 2017, 103, 133–142. [CrossRef]
Sensors 2021, 21, 3910 21 of 21

21. Sudrajat, M.I.; Moonen, N.; Bergsma, H.; Bijman, R.; Leferink, F. Evaluating Rapid Voltage Changes and its Propagation
Effect using Multipoint Measurement Technique. In Proceedings of the 2020 International Symposium on Electromagnetic
Compatibility-EMC EUROPE2020, Rome, Italy, 23–25 September 2020; pp. 4–9. [CrossRef]
22. Melo, I.D.; Pereira, J.L.R.; Variz, A.M.; Garcia, P.A.N. Harmonic state estimation for distribution networks using phasor
measurement units. Electr. Power Syst. Res. 2017, 147, 133–144. [CrossRef]
23. Khan, S.; Singh, B.; Makhija, P. A review on power quality problems and its improvement techniques. Int. J. Eng. Res. Technol.
ENCADEMS 2020 2020, 8, 1–7. [CrossRef]
24. Institute of Electrical and Electronics Engineers. IEEE Std 1159 IEEE Recommended Practice for Monitoring Electric Power Quality;
IEEE: Piscatway, NJ, USA, 2009; Volume 2009.
25. Huang, N.; Peng, H.; Cai, G.; Chen, J. Power quality disturbances feature selection and recognition using optimal multi-resolution
fast S-transform and CART algorithm. Energies 2016, 9, 927. [CrossRef]
26. Saadat, A.; Hooshmand, R.-A.; Tadayon, M. Flicker Propagation Pricing in Power Systems Using a New Short-circuit Based
Method for Determining the Flicker Transfer Coefficient. IEEE Trans. Instrum. Meas. 2020, 9456, 1–9. [CrossRef]
27. Masoum, M.A.S.; Fuchs, E.F. Introduction to Power Quality; European Copper Institute: Woluwe-Saint-Pierre, Beligum, 2015;
ISBN 9780123695369.
28. Eghtedarpour, N.; Farjah, E.; Khayatian, A. Effective Voltage Flicker Calculation Based on Multiresolution S-Transform. IEEE
Trans. Power Deliv. 2012, 27, 521–530. [CrossRef]
29. Abdelrahman, S.; Wang, M.; Milanovic, J.V.; Becirovic, E. Study of harmonic propagation in transmission networks with high
penetration of power electronics devices. In 2017 IEEE Manchester PowerTech, Powertech; IEEE: Piscatway, NJ, USA, 2017. [CrossRef]
30. Ullah, A.; Sheikh, I.U.H.; Arshad, S.; Saleem, F. Digital Active Power Filter Controller Design for Current Harmonics in Power
System. In Proceedings of the 2019 16th International Bhurban Conference on Applied Sciences and Technology (IBCAST),
Islamabad, Pakistan, 8–12 January 2019; pp. 384–388. [CrossRef]
31. Khetarpal, P.; Tripathi, M.M. A critical and comprehensive review on power quality disturbance detection and classification.
Sustain. Comput. Inform. Syst. 2020, 28, 100417. [CrossRef]
32. Jaen-Cuellar, A.Y.; Morales-Velazquez, L.; Romero-Troncoso, R.J.; Moriñigo-Sotelo, D.; Osornio-Rios, R.A. Micro-genetic algo-
rithms for detecting and classifying electric power disturbances. Neural Comput. Appl. 2017, 28, 379–392. [CrossRef]