An Intercommunication Home Energy Management System with

An Intercommunication Home Energy Management System with
Mobile Netw Appl (2012) 17:132–142
DOI 10.1007/s11036-011-0345-z
An Intercommunication Home Energy Management
System with Appliance Recognition in Home Network
Ying-Xun Lai · Joel José Puga Coelho Rodrigues ·
Yueh-Min Huang · Hong-Gang Wang · Chin-Feng Lai
Published online: 6 December 2011
© Springer Science+Business Media, LLC 2011
Abstract In present days there are wide varieties of
household electric appliances along with different
power consumption habits of consumers, making identifying electric appliances without presetting difficulty.
This paper introduces smart appliance management
system to recognize electric appliances in home networks, which uses sensing devices that measure current
to calculate the power consumption of the appliances.
The system will set the characteristics and categories of
each electric appliance, and then uses the classifications
of the electronic energy features in order to recognize
different appliances. The system searches the cluster
data while eliminating noise for recognition function-
Y.-X. Lai · Y.-M. Huang
Department of Engineering Science, National Cheng Kung
University, Taiwan Tainan, Republic of China
Y.-X. Lai
e-mail: [email protected]
Y.-M. Huang
e-mail: [email protected]
J. J. P. C. Rodrigues
Instituto de Telecomunicações, University of Beira Interior,
Beira, Portugal
e-mail: [email protected]
H.-G. Wang
Department of Electrical and Computer Engineering,
University of Massachusetts Dartmouth, 285 Old Westport
Road, North Dartmout, USA
e-mail: [email protected]
C.-F. Lai (B)
Institute of Computer Science and Information Engineering,
National Ilan University, No. 1, Sec1, Shen-lung Road,
I-Lan, 260, Taiwan, Republic of China
e-mail: [email protected]
ality and error detection mechanism or the electric
appliances using the current clustering algorithm. Afterwards the recognition are used to build a control list
of appliances on the platform to provide appliance intercommunication. Simultaneously, the household appliance automatic control services are integrated by
the system to control appliances based on userspower
consumption plans to realize a bidirectional monitoring
services. In actual experiments, the proposed system
achieves a recognition rate or 95% as well as successfully controls general household electric appliances in
home network.
Keywords smart appliance management system ·
appliance recognition · home energy management
1 Introduction
A smart grid introduces Information and Communication Technology (ICT) into the two-way communication between a power company and customers in a
power distribution system to optimize power generation, power distribution system, and power consumption [1]. However, in recent years, the concept of the
smart grid has extended to general families, and is
known as a home grid, where the power consumption
of household appliances is measured by a smart meter.
The smart meter is the bottommost component used in
general home grid for smart grid services. The recently
promoted smart meters, such as Google Power Meter or Microsoft Hohm, can show the total household
power consumption at present, but cannot show the
power consumption of each household appliance ,to say
nothing of information about the household appliances
Mobile Netw Appl (2012) 17:132–142
that are consuming power [2–4]. As a result, users
cannot further improve their power consumption habits
or avoid the use of so-called high-power electric appliances. A system that can accurately detect and recognize electric appliances is a subject worthy of study. The
paper is extended version of the paper A Smart Appliance Management System with Current Clustering
Algorithm in Home Network [5] accepted for GreenNets 2011. This study proposed an intercommunication
Home Energy Management System (HEMS) with current clustering algorithm in home network, which can
measure the household power consumption through a
current sensor, transmit the data back to the HEMS,
recognize each electric appliance, and then determine
whether it is working normally according to its staged
power consumption and various electronic energy features caused by its power sine wave intervals, so as to
avoid overloading problems arising from old or faulty
electrical appliances. However, older or large numbers of household appliances and wireless transmission
will cause power noise problems, which would result
in the inaccurate recognition of electric appliances.
Therefore, in this study, a set of current clustering
algorithm was presented to determine the cluster value
and cluster potential for measured current information.
When an abnormal value arises from the system, it is
identified as noise or an abnormal state according to
the clustering characteristics. In Section 2, we introduce
the smart meter and review related proposals in the
area of appliance recognition on HEMS. In Section 3,
we introduce the proposed system and describe system
structure and module design. The implementation of
the experimental platform and Experiments Analysis is
given in Section 4. We conclude with Section 5.
2 Related Work
This section briefly provides an outline of a smart meter
and relevant studies of electric appliance recognition,
which can help readers outside the specialty of the
article to understand the system structure.
2.1 Smart Meter
The smart meter is an advanced gauging instrument.
According to its design concepts, besides measuring
power consumption, some can identify electric appliances and communicate with other electronic equipments. Some smart meters are mounted with displays
to show current power consumption and corresponding
price. One type of smart meter has extended sockets [6–8],
meaning that it has one or multiple sockets and contains
voltage and current sensors. Cho et al. [6] designed
a Smart Multi-Power Tap (SMTP) for extension-line
smart sockets to obtain the position information of a
smart socket, thus, preparing for subsequent situations
of sensor systems. Park et al. [8] predicted and reduced
the amount of data for a smart meter in order to
reduce the load of data transmission, and verified the
accuracy rate. Yingcong et al. [9] designed and analyzed
measured information for a low-cost logic circuit with
a microprocessor to read electronic energy information, and switch off standby electric appliances to save
2.2 Appliance Recognition on HEMS
The HEMS is combined with a smart meter and related
technologies for adjusting home power to save energy
which can be integrated into a Smart Grid [10–13].
Jahn et al. [14] managed and controlled electric energy
information of electrical appliances using Hydra Middleware, that allow users to identify electric appliances
and obtain power utilization information directly from
home through intelligent mobile phones with image
recognition. Son [15] used PLC (Power Line Communication) to build HEMS, which used a smart meter to
monitor the measured power consumption and notifies
users through a network for remote monitoring, as well
as enabling power utilization planning according to user
demands and power rates. The present HEMS systems
has a defect in household appliance control, since it is
difficult to position every electric appliance being used
and apply control to specified appliances. Ito et al. [16]
designed special electric energy parameters to analyze
voltage and current wave signals, indicated that, realize
effective recognition and instance some simpler calculated parameters. Ruzzelli et al. [17] build a RECAP
(RECognition of electrical Appliances and Profileing
in real-time) system that identifies specific electric appliances by creating electronic energy feature parameters that were stored in the database and a neural
algorithm was used for recognition and displaying on
the user interface. Akbar et al. [18] used Fast Fouriere
Transform to convert the time-domain current wave
forms to frequency domain signals to obtain special
electric parameters for ease of recognition.
3 Smart Appliance Management System
This section introduces the overall system and expatiates on the various function modules. Then the power
clustering was described how to reduce the noise and
error. Finally, the appliance recognition was presented.
3.1 System Overview
Figure 1 is the system scenario consisting of smart
meters, control modules, and HEMS. The smart meter transfers the electric energy information to HEMS
through the wireless transmission interface and measures power consumption of electric appliances. HEMS
manages the energy within household appliances, integrates and transfers the power consumption information of all smart meters to Automatic Meter Reading
(ARM) or Home Information Display (HID) in a home
network for display purposes. HEMS also recognizes
appliance using Electric Energy Feature Parameters
(EEFP) captured from the power information. But
large numbers of household appliances and wireless
transmission causes power noise problems which will
affect the EEFP. Abnormal error detection rates that
are cause by rates can be reduced using current clustering.
3.2 Smart Meter Design
In this study, a smart meter measured the power consumption of the household appliances, which mainly
composed of an energy metering integrated circuit (IC),
voltage and current sampling circuits, and a microprocessor, to obtain voltage and current signals. This
study uses the energy metering ID ADE7763 chip
produced by Analog Devices which can be connect
with a variety of power measurement circuits, including current converter circuit and low resistance voltage divider circuit. The metering system start to take
42 simples in a period when voltage value equal 0
and raise state. After using an energy metering IC to
measure the power information, the microprocessor
with communication interface for external transmission receives the results, allowing administrators to
Fig. 1 The system scenario of
appliance management
Mobile Netw Appl (2012) 17:132–142
understand the measurement results remotely, monitor
household power consumption systems, or carry out
additional commands to the microprocessor. The interface communication structure is shown in Fig. 2. The
micro-controller unit (MCU) controls the relays after
powering-on, and sockets control whether to power
on. Current values from a current sensor will be converted to digital signals through a Digital-to-Analog
Converter (DAC) if needed. The digital signals are
sent to the MCU and then to load side. The MCU can
use ZigBee sensors to transmit data or commands to
central control center [19–21], or record data in the
electrically-erasable programmable read-only-memory
3.3 Definition of Electric Energy Feature Parameters
This electric appliance recognition method of this study
bases its calculations on the unique EEFP, and multiple
and special energy features tend to be obtained as
reference for identification. Hence this study designs a
data structure to store EEFP, and each parameter are
described as below:
State: Number corresponding to the name and state
of the electric appliance.
Max: Maximum current.
Min: Minimum current.
RMS: The root mean square current.
The current signal is obtained through the root
mean square (RMS) operation, and its expression
is shown in Eq. 1.
I 2 (t)dt
Mobile Netw Appl (2012) 17:132–142
Fig. 2 The design of smart
Due to the time signal sampling, Eq. 1 must be
converted to Eq. 2.
j=1 I ( j)
The process of Eq. 2 in the hardware is as follows:
after the integration of the digital signal, the multiplier obtains the square of the current signal, and
accumulates the signal through a low pass filter.
The RMS current value can then be obtained from
the square root operation.
Avg: Average current.
Power: actual power consumption
Here the instantaneous power consumption can be
calculated as per Eq. 3.
p(t) = v(t) × i(t)
H(X) = E(I(X))
V and I in Eqs. 4 and 5 are respectively the RMS
values of the voltage and current, so Eq. 3 can be
expressed as Eq. 6.
p(t) = V I − V Icos(2ωt)
H(X) =
E is the expected value function, I(X) is the selfinformation of X, and I(X) is a random variable. If
p represents the probability mass function of X, the
equation of entropy can be expressed as:
The instantaneous voltage v(t) and instantaneous
current i(t) in Eq. 3 can be expressed as Eq. 4.
v(t) = 2 × V sin(ωt)
i(t) = 2 × I sin(ωt)
Ptoa: Self-determined parameter, ratio of peak
value to average value.
Entropy: Information entropy measures the expected value of the occurrence of a random variable, and expresses the disorder of information.
The information entropy is calculated in the hope
of knowing the disorder of distribution of current
information. If the instant current value X is a random variable, its range is {X1 , · · · , Xn }, the entropy
value H is defined as:
p(xi )I(xi ) = −
p(xi ) log p(xi )
Max-count and Min-count: The power factor is
difficult to calculate due to the complexity of current electric appliances, this study determines the
difference between maximum and minimum current,
as based on the signal triggered by the voltage
square wave rising edge, as the basis of voltage and
current offset.
For the sinusoidal waveform, the actual power can
be obtained through the instantaneous power, as
shown in Eq. 7.
p(t)dt = V I
3.4 Recognition Transmission Packet Unit
This section describes the recognition transmission
packet unit format and corresponding function. The
smart meter transfers information through PLC or
wireless networks for HEMS for recognition of the
transmission packet. The packet can be seen in Fig. 3,
the start of the frame is represented by the start bit. The
head block defines the signal sending device (Smart
Meter) position and signal receiving device (HEMS)
position, and contains the initiator local position of
4bits, the destination local position of 4bits, End of
Message 1bit (EOM), and Acknowledge (ACK). The
data block represents 8 bits of information data, EOM,
and ACK, including explanatory information and the
required electric energy feature parameters.
In the transmission function, the State_request function is an application layer function to transmit data
by requesting from the network layer. State length
is the length of data to be transferred; Ptr_feature is
the structure pointer of the status data, (the detailed
status data structure will be shown in the next section); Currentstate is current state number; Txoptions
is option of the transmission mode, including whether
to broadcast or not, transmission address, and security mechanism. State_indication is the network layer
function that transfers to the target application layer,
other than transmitted parameters; there are transmission quality parameters and transmission flags to
identify the accuracy of the transmission. State_confirm
is the network layer signal sent back to the application
layer, indicating whether if the transfer of status data is
Fig. 3 The recognition
packet format
Mobile Netw Appl (2012) 17:132–142
3.5 Power Clustering
When the voltage and current information is received,
the voltage information must be normalized to 110 V
first. During wireless transmission, the incorrect values can be caused by transmission interferences and
noise effect like those shown in Fig. 4a, affecting identification accuracy of electric appliances. For electric
appliances under different operating states, conversions between its current and phase angle will be shown
as a clustering distribution, and the final clustering
distribution will be within a fixed number of regions.
Hence according to the traits mentioned, whether if the
current stays within the same state can be determined
by whether if the subsequent trace falls within the
clustering range through clustering operations. If the
value appears out of the clustering range, it must be
differentiated as abnormal clustering values or instantaneous noise distribution. Using the current clustering
algorithm, the identification rates of electric appliances
can be effectively increased while reducing abnormal
error rates caused by noise.
In this study, the subtractive clustering method [22]
in the neural algorithm processes the power clustering
characteristics. This method regards all data points as
potential center points and select clustering standards
according to the density of surrounding data points.
Mobile Netw Appl (2012) 17:132–142
The subtractive clustering method is independent of
system dimension complexity, however is proportional
to amount of data. Here, Mi is supposed to be the
power group, and ra is the influence distance of the clustering group center point and is a positive constant. The
potential value Pi (Eq. 10) of the sampling point group
Mi can be calculated, that represents the potential of
this point becoming the clustering center point.
Mi − M j2
Pi =
exp −
r 2
After the calculation of all potential values P of
the sampling points, the Mc1 with the highest potential
value is selected as the first clustering center point. The
potential values of the other points then need to be
modified, as per the following Eq. 11:
Mi − M j2
Pi = Pi − PC1 exp −
r 2
Where in, rb is a value to be set to avoid getting too
close to the last clustering center point Mc1 . It needs
to be greater than ra , and its recommended value is
1.5 times that of ra . After this process is repeated, the
sampling point group M can be divided into subgroups,
wherein, ε and ε are the upper and lower limit ratios of
the potential value, which are defined in this study as
0.5 and 0.15, respectively. The clustering current value
shows as Fig. 4b.
3.6 Appliance Recognition
The system set up a factor queue of the various eigenvalues in sequence. When data from new electric appli-
ances are generated and a factor queue of ten values
is prepared, they are input into the search system and
results are obtained [23, 24]. Then the results are saved
in the form of a database. Each element within the
queue has the same structure which includes data for
device model, device importance, device description,
and power characteristics. The power characteristics
has a structure comprised of above parameters. After, the retriever carries out corresponding database
operations from the factor queue based on different
factor properties. Different cases of factor operations
are shown below:
EEEP operation: The same power characteristics in
the data base are compared to eliminate elements with
overly different power characteristics in the database.
In this operation, a set of appliance-recognition algorithm is implemented, which is a modified algorithm
based on the hierarchical match classify model. The
assumption here is that the system regularly captures
clustering EEEP of the electric appliance as the recognition standard, and the system will save the currently
average current of the measured power clustering in
the D1 time used as EEEP for the first recognition.
The system compares EEEP of the power clustering
with a organized list obtained through the learning as
a comparison target. Electric quantity clustering data
are identified with the M1 appliances if M1 appliances
are successfully classified as being related, for the next
parameter step by step, until a complete power model
can be identified and the recognition algorithm is completed, as shown in Fig. 5.
Presume that there are M electric appliance models,
therefore it would take M ∗ T1 (D1 ) to find the electric
power models with first parameters, respectively, and
T2 (D1 ) to identify and compare the next classified
Fig. 4 The current waveform (a) in normal (b) with the power clustering algorithm
Mobile Netw Appl (2012) 17:132–142
group. The total time consumption is as shown in
Eq. 12:
T(M) = M ∗ T1 (D1 ) + M1 ∗ T2 (D1 ) + M1 M2 ∗ T3 (D1 )
+M1 M2 ∗ T3 (D1 ) + · · ·
+M1 M2 · · · M N−1 ∗ T N (D1 )
By analogy, suppose each time the same amount of
models is successfully identified, and that M1 = M2 =
· · · = n, and T2 = T3 = .. = T N it is simplified as Eq. 13:
T(M) = M ∗ T1 (D1 ) + nT2 (D1 ) ∗ (nm − 1)/n − 1 (13)
An index array will be obtained when the operation
that corresponds to each factor has been completed.
Fig. 5 Comparison programs of electric characteristics
The array records the element index in the database
and continuously inputs the structural sequence of this
device into the operating queue in sequence. The operating queue is shown in the system while electric
appliance is being used, and device models may also
be collected by continuous control commands from this
operating queue.
4 System Implementation and Analysis
This section introduced the implemented system for
the two-way recognition and control service, where the
system can identify the electric appliances and their
power consumptions, and various home appliances can
be controlled through a user interface.
Mobile Netw Appl (2012) 17:132–142
4.1 The Smart box and User Interface
randomly in identification analysis, and there were 30
experiments in each stage.
This study combines a smart meter that measures the
current and voltage data of electric appliances and
converts that data into required electric energy parameters, as shown in Fig. 6a, with a universal control
module in order to form a smart box. Then it transfers
through a wireless transmission interface to HEMS.
The control services of current electric appliances are
managed by the control module with IR and relay
set control interface, which has a wireless transmission
module for user control using wireless transmission
mode. Figure 6b shows the result interface created in
this study. The main goal of this study was to measure
the power consumption of everyday household appliances and to be able to recognize electric appliances
while permitting users to inquire related information
remotely through Internet. The household temperature
and humidity comprises the measure information received by sensors within surrounding environment that
can display power, voltage, and current information at
the same time. While content-awareness was defined
as control automation within electric appliances due
to environmental and historical user information. The
definition of user-control was that all electric appliances which are self-controlled by users.
4.2.1 Relation between recognition accuracy
and recognition time
The training and recognition time was used to base the
recognition accuracy study in this experiment, in which
the rate was defined as:
∗ 100%
which NS is number of successes for recognition and
NT is number of total test. When the training time was
lacking (15 S), as can be seen from the experimental
results, the sample establishment would not be complete, creating recognition difficulties. The recognition
rate was still difficult to improve when the recognition
time was lengthened, and if the time span was too short
recognition difficulties would still occur. The best training time was seen from experimental results to be 60 s.
The recognition accuracy was 92% when recognition
time is 120 s, and the accuracy can reach as high as 95%
(Fig. 7).
4.2.2 Relation between Recognition Accuracy
and the Current Clustering Algorithm
4.2 Experiments and Analysis
A total of 40 different household appliances were
used in this study for experimental analysis. In the
experiment, at most six electric appliances are started
The effect of the clustering algorithm on the system
recognition rate analysis is done in experiment, 0. An
experiment is tested with a training time of 60 s with a
recognition time of 90, 120, 150, and 180 s respectively.
Fig. 6 The implement of
(a) smart box included smart
meter and universal control
module and (b) the user
Fig. 7 Relation between recognition accuracy and recognition
The following figure displays the experimental results,
where it can be seen that the recognition accuracy was
around 91.5% on average when the current clustering
algorithm was used, and 79% on average without using
the algorithm, due to noise effect on data collection,
sampling and identification (Fig. 8).
4.2.3 The Traf f ic Packet Size
of Communication Service
Figure 9 shows the results of traffic pack size between
electrics and services. It mainly transmits to HEMS the
current value for recognition service. Hence, between
the packet size and number of electrical, the relation-
Mobile Netw Appl (2012) 17:132–142
Fig. 9 The Packet Size for interconnecting service with different
ship is near-linear. Downside being, 15 s are needed to
complete electrical information on recognition service,
taking 3.1 Mb packet transmission capacities for every
kind of electrical. For control configuration, the number of packets required to be transmitted is determined
by the control commands and the states. Such as, the
packet size for simple switching of the electrical is
about 1.2 Mb, while TV has larger states and control
commands but on control configuration requires 14 Mb
in packet size. Commands are usually set at 10 bytes.
For example, a simple switching of the electrical
packet size is about 1.2 Mb, but TV which has a large
states and control commands required packet size requires 14 Mb on control configuration. Commands are
fixed size at 10 bytes.
4.2.4 The Ef fect of Feature Parameters
Fig. 8 Relation between recognition accuracy and the current
clustering algorithm
In order to provide appliance recognition service, this
study proposes ten kinds of Electric Energy Feature
Parameter, and in this section the efficiency of recognition will be analyzed. Within the scope of this research,
20 different appliances are chosen to be recognized
which are classified into four categories: capacitance,
inductance, resistance, and hybrid. Each category
has influences towards appliances, and to explore the
main influences some parameters are chosen in EEFP
analysis. Figure 10 shows the results, max_count and
min_count are mainly used to find the offsets of voltage
and current of appliance that belongs to inductance,
capacitance, or resistance categories that has the same
power. The degree of data is commonly quantified
using entropy. Hybrid type appliances can easily bring
their currents to unstable states; hence the entropy
Mobile Netw Appl (2012) 17:132–142
Yueh-Min Huang’s research work in this paper is supported
by project NSC 99RC13 conducted by National Cheng Kung
University under the sponsorship of the National Science Council
Fig. 10 The effect of feature parameters for recognition
is applied to the system to the identification between
hybrid and other appliances. An important parameter
for feature recognition is standard deviation, but if
lacking the standard deviation the recognition accuracy
decreases to 72%.
5 Conclusion
In this study, a smart appliance management system
using current clustering algorithm in home network was
presented. It uses a smart meter to measure power
information and transmits data through wireless transmission back to the management platform. This enables
users to know electric appliances currently being used
and the power they consume by identifying the devices,
and users can use the control interface to remotely
control household appliances. With the aid of context information sensors and user habits it establishes
content-aware service functions and can reach a recognition rate as high as 95% with the current clustering algorithm and the establishment of identification
samples. In the future, research will be focused on
creating planning control models matching with cloud
services in order to expand the range of recognition and
retrieving identification samples.
Acknowledgements This work has been partially supported by
the Instituto de Telecomunicações, Next Generation Networks
and Applications Group (NetGNA), Portugal, and by National
Funding from the FCT Fundação para a Ciência e a Tecnologia
through the PEst-OE/EEI/LA0008/2011 Project.
Chin-Feng Lai’s research work in this paper is supported
by project NSC 100-2511-S-197-004 conducted by National Ilan
University under the sponsorship of the National Science Council
1. Cho HS, Yamazaki T, Hahn M (2009) Determining location
of appliances from multi-hop tree structures of power strip
type smart meters. IEEE Trans Consum Electron 55(4):2314–
2. Tajika Y, Saito T, Teramoto K, Oosaka N, Isshiki M (2003)
Networked home appliance system using Bluetooth technology integrating appliance control/monitoring with internet
service. IEEE Trans Consum Electron 49(4):1043–1048
3. Serra H, Correia J, Gano AJ, de Campos AM, Teixeira I
(2005) Domestic power consumption measurement and automatic home appliance detection. In: Proc. of international
workshop on intelligent signal processing, Faro, Portugal,
pp 128–132
4. Lien CH, Bai YW, Chen HC, Hung CH (2009) Home appliance energy monitoring and controlling based on power
line communication. In: Proc. of digest of technical papers
international conference on consumer electronics, Las Vegas,
NV, pp 1–2
5. Chen SY, Lu YS, Lai CF (2011) A Smart Appliance Management System with Current Clustering Algorithm in Home
Network. In: Proceedings of First ICST international conference on green communications and networking, pp 1–4
6. Cho HS, Kato T, Yamazaki T, Hahn M (2009) Simple and robust method for detecting the electric appliances using markers and programmable logic devices. In: Proceedings of the
IEEE 13th international symposium on consumer electronic,
pp 334–338
7. Heo J, Hong CS, Kang SB, Jeon SS (2008) Design and implementation of control mechanism for standby power reduction. IEEE Trans Consum Electron 54(1):179–185
8. Park S, Kim H, Moon H, Heo J, Yoon S (2010) Concurrent simulation platform for energy-aware smart metering
systems. IEEE Trans Consum Electron 56(3):1918–1926
9. Yingcong Y, Binqiao L, Jing G, Yehui S (2010) A design of
smart Energy-saving power module. In: Proceedings of the
IEEE conference on industrial electronics and applications,
pp 898–902
10. Liu CW, Luo CC, Lin PY, Lu GC, Wu WC, Tsai JI, Hsueh
CY (2011) Develop a power quality measurement system
integrated with HAN home energy management system.
In: Proceedings of 2011 4th international conference on electric utility deregulation and restructuring and power technologies, pp 1506–1510
11. Han J, Choi CS, Park WK, Lee I (2011) Green home energy
management system through comparison of energy usage between the same kinds of home appliances. In: Proceedings of
the IEEE 15th international symposium on consumer electronics, pp 1–4
12. Rossello-Busquet A, Soler J, Dittmann L (2011) A Novel
Home Energy Management System Architecture. In: Proceedings of 2011 UkSim 13th international conference on
computer modelling and simulation, pp 387–392
13. Han J, Choi CS, Lee I (2011) More efficient home energy management system based on ZigBee communication
and infrared remote controls. IEEE Trans Consum Electron
14. Jahn M, Jentsch M, Prause CR, Pramudianto F, Al-Akkad A,
Reiners R (2010) The energy aware smart home. In: Proceedings of the international conference on future information
technology, pp 1–8
15. Son YS, Pulkkinen T, Moon KD, Kim C (2010) Home energy
management system based on power line communication.
IEEE Trans Consum Electron 56(3):1380–1386
16. Ito M, Uda R, Ichimura S, Tago K, Hoshi T, Matsushita Y
(2004) A method of appliance detection based on features of
power waveform. In: Proceedings of the international symposium on applications and the internet, pp 291–294
17. Ruzzelli AG, Nicolas C, Schoofs A, O’Hare GMP (2010)
Real-Time recognition and profiling of appliances through a
single electricity sensor. In: Proceedings of the IEEE communications society conference on sensor Mesh and Ad Hoc
communications and networks, pp 1–9
18. Akbar M, Khan DZA (2007) Modified nonintrusive appliance load monitoring for nonlinear devices. In: Proceedings
of the IEEE international multitopic conference, pp 1–5
Mobile Netw Appl (2012) 17:132–142
19. Lin WD, Lai CF, Ke CH, Cheng RS (2010) OSGi-Based intelligent context-aware middleware for smart home appliances.
Journal of Internet Technology 11(7):935–941
20. Min C, Gonzalez S, Leung V, Qian Z, Ming L (2010) A
2G-RFID-based E-healthcare system. IEEE Wirel Commun
21. Park WK, Han I, Park KR (2007) ZigBee based dynamic
control scheme for multiple legacy IR controllable digital consumer devices. IEEE Trans Consum Electron 53(1):172–177
22. Chiu SL (1994) Fuzzy model identification based on cluster
estimation. J Intell Fuzzy Syst 2:267–278
23. Lin GY, Lee SC, Hsu JYJ, Jih WR (2010) Applying power
meters for appliance recognition on the electric panel. In:
Proceeding of the 5th IEEE conference on industrial electronics and applications, pp 2254–2259
24. Guvensan MA, Taysi ZC (2010) Environmental sound classification for recognition of house appliances. In: Proceeding
of IEEE 18th signal processing and communications applications conference, pp 431–434
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF