Content uploaded by Marius Marcu
Author content
All content in this area was uploaded by Marius Marcu on Sep 15, 2015
Content may be subject to copyright.
Advances in Electrical and Computer Engineering Volume 15, Number 1, 2015
Anomaly Detection Using Power Signature of
Consumer Electrical Devices
Cosmin CERNAZANU-GLAVAN, Marius MARCU
Politehnica University of Timisoara, Timisoara, 300006, Romania
cosmin.cernazanu@cs.upt.ro, marius.marcu@cs.upt.ro
1Abstract—The use of the smart grid for developing
intelligent applications is a current trend of great importance.
One advantage lies in the possibility of direct monitoring of all
devices connected to the electrical network in order to prevent
possible malfunctions. Therefore, this paper proposes a method
for an automatic detection of the malfunctioning of low-
intelligence consumer electrical devices. Malfunctioning means
any deviation of a household device from its normal operating
schedule. The method is based on a comparison technique,
consisting in the correlation between the current power
signature of a device and an ideal signature (the standard
signature provided by the manufacturer). The first step of this
method is to achieve a simplified form of power signature
which keeps all the original features. Further, the signal is
segmented based on the data provided by an event detection
algorithm (values of the first derivatives) and each resulting
component is approximated using a regression function. The
final step consists of an analysis based on the correlation
between the computed regression coefficients and the
coefficients of the standard signal. Following this analysis all
the differences are classified as a malfunctioning of the
analyzed device.
Index Terms—Feature extraction, Pattern matching, Signal
analysis, Signal processing.
I. INTRODUCTION
The increasing of the electrical network over time and the
fact that it is broadly interconnected were the main reasons
for the occurrence of what it is called now a smart grid. The
difference between a power network and a smart grid is that
the latter uses digital technology to gather information used
to improve the services. [1] For this reason, we can say that
one of the priorities of the new type of network is to create a
link between the users and the providers of electric power.
This new infrastructure allows both utility providers and
third party companies to develop various applications which
provide specific services to their customers [2].
These applications offer numerous possibilities of
analysis of the operating mode of the power network and
permit better control of energy consumption costs. Future
consumer devices may be smart enough to communicate
with the grid infrastructure and metering devices, describe
their energy needs and select the optimum usage pattern
based on the best available tariff and technical assistance
and diagnosis (for this article).
The goal of our work is to enable customers to receive
information on the operation mode of personal electrical
devices [3]. We want to extend the smart grid to the
customer location and build a network of smart meters that
monitor all appliances.
This work has been supported by the project CHISTERA/1/01.10.2012 –
”GEMSCLAIM: GreenEr Mobile Systems by Cross LAyer Integrated
energy Management”.
Today, smart meters can analyze power consumption
variations of consumer devices and may send this
information to the utility company. The companies use data
received from their customers to generate their energy bill
and predict and optimize the energy distribution load
patterns. Instead, we propose collecting, storing and
processing data from smart meters locally and use their
power signatures for intelligent application development.
The variation in time of the power consumed by an
electrical device when executing a task is defined as the
power signature of that device. It can be seen that, in this
way, a power signature is specific to a certain device when
executing a particular task.
The method proposed in this paper is the automatic
detection of malfunctioning for low-intelligence consumer
electrical devices. Anomaly detection is made by comparing
the current power signature with an ideal signature (standard
signature) supplied by the manufacturer. Sections of the
signal that are not identical are considered malfunctioning
for the device.
Given the oscillatory nature of the signal (Fig. 2) it is
difficult to implement an automated method of comparison
based on the original signal. For this reason it is necessary to
achieve a simplified form of signal, which should retain all
the essential characteristics of the original signature (values
of the amplitude, variations of the frequency, changes of the
first derivative, etc.). For each component, the coefficient of
correlation between it and the standard components is
calculated. The automated analysis is based on these values
which form a correlation matrix.
In the next chapter we present the main achievements in
this domain, highlighting the advantages and disadvantages
for each achievement. In Chapter 3, we will make a brief
summary of the process for obtaining a simplified shape for
the power signature. This process has been already validated
and published in [4]. Chapter 4 presents the anomaly
detection algorithm using the shape signal from section 3,
together with the results of the experiments. The last section
presents the conclusions obtained in this study.
II. RELATED WORK
One of the first uses of the power signature occurred in
the PCB (printed circuits boards) industry. In order to
differentiate whether a PCB operates normally or exhibits
abnormalities, the voltage and current values were recorded
during a continuous operating mode. [5] After that, these
values (power signatures) were compared with a standard
shape, representing the normal operating mode for the
89
Digital Object Identifier 10.4316/AECE.2015.01013
1582-7445 © 2015 AECE
[Downloaded from www.aece.ro on Saturday, September 12, 2015 at 06:54:12 (UTC) by 195.49.254.25. Redistribution subject to AECE license or copyright.]
Advances in Electrical and Computer Engineering Volume 15, Number 1, 2015
boards. The resulted differences constituted a starting point
in the decision process that specifies whether there can be
operated problems with the board.
The monitoring process of the running mode can be done
with many devices on the market [6]. Of these, the most
common are Kill a Watt or Watts Up Pro (used by us to
collect the data needed).
Current researches are performed with the aid of complex
devices. Most of them are focused on the need of low
consumption of domestic users. A recent study [7] provides
an interesting solution for disaggregated consumption of
each electrical device used. The authors manage to extract
information that ultimately leads to a solution to calculate
the power consumption for each device. It should be noted
that much of the research is focused on how to obtain the
power signature, because the authors do not aim to make an
analysis process.
Starting from this, we must mention that there are
numerous solutions to obtain the power signature. One of
them is even suggested by the authors [8] and consists in an
intelligent framework based on wireless technology for a
continuous monitoring. Thus, by using a network of sensors,
it is provided a remote monitoring for any type of building:
from buildings to offices and private homes (Fig.1)
A new solution is proposed in [9] and uses the new
concept of Internet of Things (IoT) in order to speed up the
process of data acquisition. In this way, the authors ensure
us that it can prevent certain natural disasters and also, the
transmission of information is done with a low cost.
Subsequent to the monitoring and acquisition of the
power signature processes there is another process which
tries to analyze it. Thus, it ensures the extraction and
processing of important information available in the power
signature.
Figure 1. Overall smart monitoring infrastructure
This process takes place in two parts. In the first part it is
obtained a simplified form of signal that contains all the
parameters of the original one. In the second part, the
obtained signature form is analyzed and processed, and the
results are provided.
The method of obtaining the simplified form of the power
signature is well presented in [4] and it will be summarized
in the next chapter.
There are many methods for analyzing the power
signature and extracting the important information from it.
We start by presenting a method for classifying domestic
devices using two features: the power signature and the
harmonic features. [10] This study attempts to solve the
problem of how to predict the power consumption especially
for residential housing. For this purpose, it determines
several categories of consumer and tries to classify every
device in a certain category. A positive outcome of this
approach lies in a better understanding of the power
co
tained by the
m
no automatic power signature
an
mprove the results and
to extend the existing methodology.
lished and validated by
the authors and can be found in [4].
nsumption by the residents.
This article seeks to determine how a power signature is
changed when there appears some malfunctioning of the
electrical device. This is done by comparing the current
power signature with an ideal signature (ob
anufacturer in the normal operation mode).
Similar goals had as well the projects described in [11]
and [12]. The research in [11] focused on two main
directions: (1) the identification of a method capable to
determine the load characteristics of a device and (2) several
ways of signal processing needed to determine various types
of workload for the analyzed device. The authors of [11]
used the AC voltage and current signals and their detailed
variations. In our work we used the high level digitized
measured values of active power, voltage and current
consumption. The authors of [12] explain the method by
which there was implemented a solution used for monitoring
the AC (air conditioning) devices. Using a sensor network, it
is controlling several AC devices in different types of
buildings. The paper is focused on the monitoring solution
presentation that is mainly used for energy usage reporting
and analysis. However, some power signatures of daily
usage patterns for several office and consumer devices are
presented and discussed. But
alysis is presented in [12].
An interesting analysis [13], based on power signature is
made in order to detect the type of fault which can appear
into an induction motor drive. The analysis attempts to
separate the electrical faults cases from mechanical faults.
Detecting faults in an electrical device, using different
signatures (e.g. load signatures or frequency response
analysis signatures) was also the goal for the authors from
[14] and [15]. A practical approach of using power signature
and power factor was already done by the authors in [16].
However, in this paper we want to i
III. SIMPLIFIED FORM OF SIGNAL
In this chapter we will briefly present how to obtain a
clear form of a power signature. A more detailed description
of the process has already been pub
Figure 2. The power signature of a washing machine running a normal
rogram
p
90
[Downloaded from www.aece.ro on Saturday, September 12, 2015 at 06:54:12 (UTC) by 195.49.254.25. Redistribution subject to AECE license or copyright.]
Advances in Electrical and Computer Engineering Volume 15, Number 1, 2015
A. Filtering the signal with a low-pass filter
Original signal (Fig. 2) has many oscillations due to how
the power device turns on / off various electrical modules. In
our case, for a washing machine, the electric motor is turned
on / off every few seconds.
To eliminate these oscillations, the first method of signal
processing is to filter it by using a low-pass filter. The signal
obtained after filtering has a clearer shape and can be more
easily used, for further processing, than the original one. His
form remained the same and the amplitude of oscillations
was much reduced. For further reduction of oscillations in
the next step we will apply the Fast Fourier Transform
(FFT) to this signal.
B. Applying a Fast Fourier Transform
The FFT method transforms the signal from the time
domain into the frequency domain. This method aims at
selecting a fixed number of frequencies that are part of the
signal and tries rebuilding it using only these frequencies.
To reconstruct the signal, the most important frequencies
must be selected together with their amplitude. The manual
selection of these has led to poor results and that is why we
created an automatic method to select N number of these
frequencies (in descending order of their amplitude values).
After this transformation, the disappearance of some
oscillations that do not contain relevant information for the
signal can be observed. The FFT method simplifies the
signal shape but keeps constant all the values for the main
components of the signal.
C. Smoothing functions
Into a last attempt, the following processing method tries
to reduce more some of the remaining oscillations. This was
achieved by using a smoothing function for the oscillations
which presented large amplitudes. In our case, there are
several smoothing functions that can be applied. The best
results were obtained by applying a Gaussian smoothing
function and these can be seen in Fig. 3
After applying this method the signal has the best shape
obtained so far. This shape will be used extensively in the
next chapter to make a comparison between signals.
D. Extraction of other parameters
From the current signal shape there must be extracted
more parameters that will help later in the analysis stage.
These parameters will be extracted using two methods:
power spectral density (PSD) and event detection algorithm.
Figure 3. Event detection algorithm applied to our signal
An event detection algorithm is necessary when we
search certain pieces of a signal or when we try to segment
the signal in multiple parts (our case). Detection algorithm
was built by calculating the first derivative for our signal.
This algorithm finds all the behavioural changes that occur
in signal evolution. Thus, in our case, we can see that the
algorithm clearly showed the beginning / end of each event
(Fig. 3).
In the next chapter we will analyze the signal shape
together with the rest of the parameters and we determine
the signal dynamics and detect its anomalies.
IV. DYNAMIC ANALYSIS OF THE SIGNAL
An analysis of the entire shape of signal obtained in the
previous chapter is difficult for an anomaly detection
algorithm using correlation. Because of that, we decided to
split the obtained shape into pieces (components) that can be
analyzed more easily. Furthermore, such segmentation
provides us a better localization of anomalies.
A. Signal segmentation
Signal segmentation is done based on the signal obtained
from the event detection algorithm. A simplistic approach in
this case could be: any non-zero value of the signal could be
seen as a change in the behaviour of the device. The higher
the value, the greater change has occurred in signal
evolution in that moment.
This approach cannot be done in our case because the
signal shows continuous oscillations and that is why most of
the values are non-zero. Analyzing the values obtained from
event detection algorithm, we chose a cutting threshold that
separates a neutral area (values between [-2, 2]) for an area
of interest for us. Using this threshold, the important
components of the signal can be seen in Fig. 4.
Figure 4. Detection of signal components using an algorithm based on
values from event detection method after applying a threshold cut
Figure 5. Detection of short time intervals where signal has a transient
behaviour (anomalies)
These components succeed segmentation for a power
signature signal into constituent parts. Also, the algorithm
91
[Downloaded from www.aece.ro on Saturday, September 12, 2015 at 06:54:12 (UTC) by 195.49.254.25. Redistribution subject to AECE license or copyright.]
Advances in Electrical and Computer Engineering Volume 15, Number 1, 2015
manages to detect other short time intervals too, in which
signal has a transient behaviour (Fig. 5). It can be observed
the presence of these intervals in the beginning/end of the
whole program and when rinsing cycle starts. Because these
intervals have a short period of time, an automated analysis
of them could not be done, so they were analyzed in a
separate mode.
The other proposed segmentation method is based on
normalization of a signal according to another signal. For
this method we need a previous processed signal (standard
signal - considered to be a normal behaviour of the device).
This signal must have segmentation intervals defined in
advance by the user. Further, the normalization operation is
performed in the time domain for our signal and after that,
the segmentation operation is applied to time intervals
obtained from the standard signal.
Figure 6. Segmentation using a standard signal for other two different
signals
Figure 7. Signal component reconstruction using polynomial regression
In Fig. 6 we can see the segmentation result for two
signals using a standard signal. Some components have
failed to be detected and for others the beginning / end are
totally wrong.
The results vary from signal to signal and that is why we
propose this method to be used only as a secondary one.
B. Regression function for signal components
Defining a regression function for each signal component
is an important part of our approach because an analysis of
the signal is much easier to be done in this case. Thus, a
function is characterized by a reduced set of values
(parameters) than a relatively large number of sample
signals. For our signal, we try finding regression functions
from three classes: polynomial, exponential and a
combination of sinuses.
Figure 8. Signal component reconstruction using exponential regression
Figure 9. Signal component reconstruction using combination of sinuses
For a polynomial regression variant we chose a
92
[Downloaded from www.aece.ro on Saturday, September 12, 2015 at 06:54:12 (UTC) by 195.49.254.25. Redistribution subject to AECE license or copyright.]
Advances in Electrical and Computer Engineering Volume 15, Number 1, 2015
polynomial of nine degree. Using a polynomial of degree
greater than nine is difficult and leads to obtaining
coefficients with values less than 10-5. Fig. 7 presents the
reconstruction of some of the signal components where
polynomial regression functions are used. It can be seen that
some signal components cannot be exactly reconstructed
using polynomial functions.
For the second method we used multiple regression
exponential functions. The obtained results for the same
components of the signal are shown in Fig. 8. In this case,
the components of the signal were reconstructed even worse
(component 3 could not be reconstructed).
The last regression method uses a combination of sinuses
(Figure 9). It can be seen from Table I that using the
combination of sinuses we obtain the best results (e.g. the
smallest error). Further we used this method to find the
regression coefficients for each component of the signal.
TABLE I. SUM OF SQUARES DUE TO ERROR FOR EACH REGRESSION METHOD
Regression
method C1 C
2 C
3
Polynomial 0.1726 564.01 4.6*106
Exponential 358.61 776.72 1.5*105
Combination of
sinuses 268.33 3.63*105 6.7*107
C. Calculation of correlation indices between two
components
Correlation is a statistical measure that determines the
dependencies between two data sets and from our point of
view; it has an important role as it highlights the relations of
two signals. The correlation method is based on Pearson
product-moment correlation coefficient and it is named
"Pearson's Correlation". If we define two sets X and Y, the
correlation coefficient ρX,Y is given by (1).
,
cov( , ) XY
XY XY XY
EX Y
XY
(1)
To calculate the correlation coefficients we need two data
sets of equal length. In order to do that we should determine
the signal component that has the largest length (the total
number of contained points). For all the remaining
components will be added(at the end) multiple zero values
until we get a vector of length equal to the determined
maximum length. Thus, each component will be composed
of a number of samples equal to the maximum length.
Further, we present a correlation analysis based on three
signals (see Signal 1, Signal 2 and Signal 3 from Fig. 10)
Each of these three signals will be compared with the signal
presented throughout this paper named the standard signal.
Each component of a signal will be compared with each
of the components of the standard signal. A correlation
index between the two components will be calculated for
each comparison.
For example when we compared the Signal 1 with the
Standard signal, actually we compared the 7 components
found in Signal 1 with the 6 existing components in
Standard signal (Figure 11). The correlation matrix resulted
after comparison operation can be seen in Table II. We
noted by SCi , the i component of standard signal and Cj the
j component of the Signal 1.
Figure 10. Signals used in correlation analysis. From top to bottom: Signal
1, Signal 2 and Signal 3
Figure 11. Top: Standard signal contains SC1 - SC 6 components. Bottom:
Signal 1 and C1 - C 7 components
TABLE II CORRELATION MATRIX
SC1 SC 2 SC 3 SC 4 SC5 SC6
C1 0.983 0.315 0.685 0.330 0.030 -0.235
C2 0.324 1.000 0.159 0.468 0.169 -0.088
C3 0.925 0.315 0.811 0.349 0.064 -0.259
C4 0.479 0.600 0.469 0.548 0.697 -0.490
C5 0.489 0.611 0.483 0.741 0.526 -0.528
C6 0.441 0.592 0.397 0.871 0.279 -0.410
C7 0.829 0.487 0.881 0.412 0.117 -0.269
The last part of this chapter deals with the analysis of the
values from correlation matrix and explains how an anomaly
or a malfunctioning is detected.
D. Correlation analysis. Determining similarity and
anomaly detection
It can be seen from Fig. 11 that Signal 1 is very similar to
the Standard signal. This is also revealed by the matrix of
correlation coefficients. Thus:
93
[Downloaded from www.aece.ro on Saturday, September 12, 2015 at 06:54:12 (UTC) by 195.49.254.25. Redistribution subject to AECE license or copyright.]
Advances in Electrical and Computer Engineering Volume 15, Number 1, 2015
94
SC6.
Component C1 and C2 from Signal 1 are very
similar with the SC1 and SC2 components from Standard
signal (0.983 respectively 1.000). That means that the
washing machine works fine in the first part of the program
and does not show any anomalies.
Component C3 from Signal 1 is very similar to the
component SC1 from Standard signal (0.952 similarity) but
also with component SC3 from Standard signal (0.811
similarity). This is due to the fact that the C3 from Signal 1
has a double length (in time) than the similar component
from the Standard signal. Again, we can conclude that there
are no anomalies present.
Component C4 from Signal 1 has similarity with
the component SC5 from Standard signal (0.697 similarity).
This means that a slightly wear of the machine is present in
this part of the program.
Component C5 from Signal 1 has much similarity
with the component SC5 from Standard signal (0.741
similarity). Again, we can mention a slightly wear of the
machine for this part of the program.
Component C6 from Signal 1 is very similar to the
component SC4 from Standard signal (0.871 similarity).
Because all the SC4, SC5 and SC6 are part of the rinse
cycles, there is not a big issue, that this component is most
similar to a component other than
Component C7 from Signal 1 is very similar to the
component SC3 from Standard signal (0.881 similarity).
This is an anomaly because the two components are part of
different washing cycles. The first is from the rinse cycle
and the last is from the wash cycles.
Without C7 we have a similarity value of 0.85 between
Signal 1 and the Standard signal. That means a strong
correlation between the two signals.
An analysis for the Signal 2 showed a correlation of 0.823
between the two signals. It also revealed the existence of
another component (similar with C3) in this signal. This is
also an anomaly and can be observed in Figure 10.
The correlation for Signal 3 is only 0.793, so there are
many similarities between the two signals (there is a good
correlation). Also the analysis detected that the C3 for the
Standard signal does not have a counterpart component in
Signal 3. The absence of a standard component represents
also an anomaly.
V. CONCLUSIONS
The paper presents a method for an automatic detection of
malfunctioning for low-intelligence consumer electrical
devices. The method is fully automated and it is based on
the analysis of the correlation matrix between the
components of the signal. It managed to analyze and to
determine the degree of similarity and, to find differences
between two signals.
The method was tested on different power signatures
obtained from 10 washing machines. During testing phase,
the washing machines executed both a normal program and
fractions of program. The method captured the similar
phases with a normal program and prompted when abnormal
behaviour was present: a phase from program was skipped
or presented malfunctioning.
Also, based on similarity coefficients it is possible to
compute the degree of wear.
During the correlation analysis we showed how an
anomaly can be captured and analyzed based on the two
existing signals.
We want to further improve this method by refining the
analysis part of the correlation matrix and by taking other
parameters into consideration.
REFERENCES
[1] R. Kazman, L. Bass, J. Ivers, and G. A. Moreno, “Architecture
Evaluation without an Architecture: Experience with the Smart Grid”,
33rd International Conference of Software Engineering, May 2011.
[2] C. Laughman, K. Lee, R. Cox, S. Shaw, S. Leeb, L. Norford and P.
Armstrong, “Power Signature Analysis”, Power and Energy
Magazine, IEEE, Vol. 1, pp.56-63, 2003.
[3] I. Cepa, Z. Kocur, Z. Muller, “Migration of the IT Technologies to the
Smart Grids”, ELEKTRONIKA IR ELEKTROTECHNIKA, pp.123-
128, Issue 7, 2012
[4] M. Marcu and C. Cernazanu, “Dynamic Analysis of Electronic
Devices' Power Signatures”, International Instrumentation and
Measurement Technology Conference, I2MTC 2012, Graz, Austria,
May 2012.
[5] I. C. Miller, “IDDQ testing in deep submicron integrated circuits”,
Proceedings of International Test Conference, ITC 1999, Atlantic
City, USA, Sep. 1999.
[6] K. Fehrenbacher, “10 Monitoring Tools Bringing Smart Energy
Home”, Business Week, Apr. 2009,
http://www.businessweek.com/technology/
content/apr2009/tc20090414446611.htm.
[7] J. Froehlich, E. Larson, S. Gupta, G. Cohn, M. S. Reynolds, and S. N.
Patel, “Disaggregated End-Use Energy Sensing for the Smart Grid”,
IEEE Pervasive Computing, Special Issue on Smart Energy Systems,
Jan–Mar 2011.
[8] M. Marcu, C. Stangaciu, A. Topirceanu, D. Volcinschi, and V.
Stangaciu, “Wireless Sensors Solution for Energy Monitoring,
Analyzing, Controlling and Predicting”, Lecture Notes of the Institute
for Computer Sciences, Social Informatics and Telecommunications
Engineering, Volume 57, 2011.
[9] Q. Ou, Y. Zhen, X. Li, Y. Zhang, L. Zeng; , “Application of Internet
of Things in Smart Grid Power Transmission”, Mobile, Ubiquitous,
and Intelligent Computing (MUSIC), 2012 Third FTRA International
Conference on , pp.96-100, 26-28 June 2012
[10] S. J. Huang, C. T. Hsieh, L. K. Kuo, C. W. Lin, C. W. Chang, S. A.
Fang, “Classification of home appliance electricity consumption using
power signature and harmonic features”, Power Electronics and Drive
Systems (PEDS), 2011 IEEE Ninth International Conference on ,
pp.596-599, 5-8 Dec. 2011
[11] W. K. Lee, G. S. K. Fung, H. Y. Lam, F. H. Y. Chan, and M. Lucente,
“Exploration on Load Signatures”, International Conference on
Electrical Engineering, ICEE 2004, Sapporo, Japan, Jul. 2012.
[12] X. Jiang, S. Dawson-Haggerty, P. Dutta, and D. Culler, “Design and
Implementation of a High-Fidelity AC Metering Network”, The 8th
ACM/IEEE International Conference on Information Processing in
Sensor Networks, IPSN’09, 2009, San Francisco, California, USA.
[13] M. Drif, A. J. M. Cardoso, “Stator Fault Diagnostics in Squirrel Cage
Three-Phase Induction Motor Drives Using the Instantaneous Active
and Reactive Power Signature Analyses”, Industrial Informatics,
IEEE Transactions on , vol.10, no.2, pp.1348-1360, May 2014
[14] T. Hassan, F. Javed, N. Arshad, “An Empirical Investigation of V-I
Trajectory Based Load Signatures for Non-Intrusive Load
Monitoring”, Smart Grid, IEEE Transactions on, vol.5, no.2, pp. 870-
878, March 2014
[15] A. Abu-Siada, N. Hashemnia, S. Islam, M. Masoum, “Understanding
power transformer frequency response analysis signatures”, Electrical
Insulation Magazine, IEEE , vol.29, no.3, pp. 48-56, May 2013
[16] M. Marcu, C. Cernazanu, “Applications of Smart Metering and Home
Appliances' Power Signatures”, Instrumentation and Measurement
Technology Conference (I2MTC) Proceedings, 2014 IEEE
International , vol., no., pp. 331-335, 12-15 May 2014
[Downloaded from www.aece.ro on Saturday, September 12, 2015 at 06:54:12 (UTC) by 195.49.254.25. Redistribution subject to AECE license or copyright.]
