Matlab/Simulink Modeling and Simulation of

Electric Appliances Based on their Actual

Current Waveforms
Adel Gastli1, Senior Member, IEEE, Serkan Kiranyaz1, Senior Member, IEEE, Ridha Hamila1, Senior Member,
IEEE, Omar Ellabban2, Senior Member, IEEE
1
Electrical Engineering Department at Qatar University, P.O. Box 2713, Doha, Qatar
2
Iberdrola Innovation Middle East, Qatar Science & Technology Park, Doha, Qatar
adel.gastli@qu.edu.qa, mkiranyaz@qu.edu.qa, hamila@qu.edu.qa, oellabban@iberdrola.com

Abstract—This paper presents a novel modeling technique of time. In order to achieve a certain level of accuracy and
electric appliances using Matlab/Simulink based on their actual reliability, there is a mandatory requirement of detailed and
measured current waveforms. Home appliances were used as the instantaneous knowledge of the per-load electricity
study case, but the proposed approach can be applied to any consumption, and as a result in a typical residential house, this
electric appliance as long as the supply voltage is maintained requires a large-scale sensory tools connected to dozens, or
constant. In the proposed method, the measured current
even hundreds of individual loads [6]. In order to address these
waveform is split into two parts: transient and steady state. Each
part is stored in one data vector. The transient is stored in a long deficiencies, Non-Intrusive Load Monitoring (NILM)
vector while the steady state is represented by one cycle only (e.g. techniques have been proposed which use the data to infer the
20 ms for 50 Hz). When the appliance is switched on, the electricity usage for individual loads [7] or building
transient data vector is used during the transient period only and occupancy patterns [8]-[10]. With the data acquired and
then the steady-state data vector is repeated every supply cycle cumulated in time, certain control policies were able to be
indefinitely until the appliance is switched off or the simulation developed for trimming the peak demand by a proper
is terminated. Compared to previously published methods, the scheduling mechanism which can in particular control, for
proposed method is much more simple and accurate since it is example, heaters and air conditioning systems with flexible
based on the actual current waveform and not on any
background loads [10].
mathematical approximation or curve fitting. Finally, the
created library of models in this study will be very useful for In particular, the “on/off” model is one of the most
researchers when designing energy management. common approach, which can turn the usage of electricity in
a house in a state machine where the transition from one state
Index Terms-- Current signature, energy management, home to another occur when a device is turned on or off. The major
appliances, Matlab, modeling, simulation, Simulink, smart grids
advantage in this model is the ability to capture even low-
power loads such as bulbs, chargers, etc. The most common
I. INTRODUCTION
technique to capture such transitions in a dynamic state-space
Energy management in buildings including houses has system is Hidden Markov Models [11] that can be trained by
become a serious concern and research focus of many the Viterbi’s algorithm [12], [13] so as the seek the state of
researchers and developers of power systems. Indeed, the each load. The main disadvantage is that the number of states
demand for electricity has grown tremendously and is grows exponentially with respect to the number of loads, i.e.,
expected to grow even further during the next few decades as 2n for n loads. Despite of their simplicity, such models are also
illustrated in Fig. 1[1]. To meet the future electricity demand, faced with crucial challenges such as dynamic characteristics
it is important to develop ways of managing electricity of various appliances such as TVs, computers, monitors, etc.
consumption more efficiently and save energy through This again brings the problem of reliability and robustness
Demand Side Management (DSM). issues on surface.

This will enable matching the available energy resources

with the electricity demand and reduce CO2 footprint.
According to Fig. 1, it is clear that residential CO2 emission
intensity is similar to that of commercial end-users, and both
are higher than the industrial sector but less than the
transportation sector. Therefore, several researches were Transportation
conducted on DSM and Building Energy Management Commercial
Residential
Systems (BEMS) for residential buildings aiming at reducing Industrial
the energy consumption, hence, improving the CO2 footprint.
Although there are various methods proposed in the past
[2] - [5], there still remains severe limitations and drawbacks
such as lack of utmost accuracy and reliability, expensive and Fig. 1 Outlook of electricity generation and CO2 intensity by end-use sector
invasive tool-set and inability to sense the electricity in real- [1].

Authorized licensed use limited to: Auckland University of Technology. Downloaded on May 28,2020 at 08:45:43 UTC from IEEE Xplore. Restrictions apply.
 This study proposes a new simple yet accurate technique It is required to have two main functions to implement first
for modeling electric appliances in Matlab/Simulink using the transient waveform of the current during the transient
actual current waveforms measured directly from several period of time (Ttrans) and then the steady-state waveform for
home appliances. Even though, the focus is drawn on home one cycle (1/f) and make it repetitive for multiple cycles as
appliances, the proposed modeling approach can be applied to long as the appliance is on. Therefore, we can use two
any electric appliance. Most importantly, the actual current counters: one for the transient period and the second one for
waveforms of the appliances are acquired and collected in a the steady-state repetitive cycle. Fig. 3 shows the Simulink
library for research purposes. block diagram that implements any home appliance.
This rest of the paper is organized as follows: Section 2 SwitchON
1 Inc Cnt
presents the dataset of home appliances’ current waveforms Count
Up u y 1
boolean NOT
used in this study; Section 3 describes the modeling Rst Hit
Counter Transient k2-k1
Current

methodology and its implementation with Matlab/Simulink; CurrWavTrans_AC

Section 4 validates the models through simulation and S Q Inc

Count
Cnt
R !Q Up
comparison with actual data. Section 5 concludes the paper Rst Hit CurrWavSteady_AC
Counter Steady State
and suggests topics for future research.
OR
II. MODELING APPROACH
A. General Principle
Fig. 3 Simulink implementation block diagram for home appliances.
Appliances are modeled based on their actual current
response during switching on period and steady state The main input to this block is the switching on triggering
operation. Fig. 2 shows an example of current response during command, which takes the values 0 or 1 during off or on
switching on one home appliance (TV). states, respectively. The command is used to trigger a counter
that is programmed to count up using the same sampling time
period as the real current waveform. The counter sends its
counts to a Matlab function for analysis. The maximum counts
One steady-
State cycle
are set to the transient time Ttrans. When the counter reaches
Ttrans, it sends a triggering pulse to the Flip-Flop block, which
changes its states from zero to one. The output of the Flip-Flop
block triggers the steady state counter, which starts counting
until it reaches the 1/f. At this time, it sends a pulse on its Hit
pin that is connected to its reset pin in order to reset this
counter to zero and starts counting again until 1/f. this is
repeated indefinitely until the switch input is set to zero which
resets and disables both counters. The same cycle is repeated
when a new switching on command triggers the first counter
Transient Time: Ttrans
again. The setting of the time counters and the data store
memory Simulink block are illustrated in Fig. 4 and Fig. 5.
The outputs of both counters are input to a Matlab function
Fig. 2 Sample of current signature of home appliance during transient and in addition to the number of current samples during the
steady-state periods. transient period designated by (k2-k1), where k1 is the current
The current waveform can be split into 2 vectors: transient sample number from the measured waveform at time of
and steady state. It is recommended to consider a relatively switching on and k2 is the sample number after Ttrans
long transient period (Ttrans) so as to make sure that it includes (measured after turning on time).
all possible transient behavior of the current waveform. Fig. 6 shows the code used in the Matlab function block,
Considering that during the steady state the current waveform which generates the simulated current waveform of the
is periodic, we can pick only one cycle (1/f) waveform and appliance based on its switching state. The Matlab function
duplicate it over multiple cycles as long as the appliance code uses two vectors “CurrWavTrans” and
remains switched on. Note that f is the frequency of the main “CurrWavSteady” that are stored in memory and outputs the
supply voltage. appropriate transient and steady-state current sample,
We can base our modeling principle on the splitting respectively, based on the counters output values. Indeed, if
technique of the actual current waveform and develop the first counter output is zero which means that the switch is
Matlab/Simulink library for any home appliance as long as we off, the function outputs zero. However, when the output of
have a recorded current waveform during its switching on counter 1 is not zero but smaller than k2-k1, then it output the
operation. The advantage of this approach is that, it does not current sample from the transient waveform. When counter 1
require any curve fitting or complex mathematical exceed the maximum value corresponding to Ttrans, it outputs
approximation and representation of the current waveform. the steady-state current sample based on counter 2 output.
B. Matlab/Simulink Modeling Note that “CurrWavTrans” and “CurrWavSteady” are
To model any appliance and make it available in the library defined as global variables that are stored in data Store
of Matlab/Simulink for possible use in various applications, Memory block so that they can be known within the Matlab
the following procedure can be followed. function while they are defined outside this function. To be

Authorized licensed use limited to: Auckland University of Technology. Downloaded on May 28,2020 at 08:45:43 UTC from IEEE Xplore. Restrictions apply.
able to do so, we need to use the “Ports and Data Manager”
setting tool (see Fig. 7) that can be displayed by clicking on
“Edit Data” under the Matlab function editor upper tabs. In
this setting tool, we need to add the two data vectors and
define then as Data Sore Memory under scope tab (see Fig. 7).
The whole block system in Fig. 3 is grouped under a
Simulink sub-system as shown in Fig. 8, which is masked and
initialized with the parameters that related to considered
appliance.

(a) Transient time counter block setting

Fig. 5 Data Store Memory block setting.

(b) Steady-state time counter block setting Fig. 6 Coding of the Matlab function that generates the simulated current
waveform.
Fig. 4 Setting of time counter: (a) Transient, (b) Steady State.

Authorized licensed use limited to: Auckland University of Technology. Downloaded on May 28,2020 at 08:45:43 UTC from IEEE Xplore. Restrictions apply.
 Fig. 7 Setting of current waveform stored data for use within the Matlab
function.

(a) Parameters and Dialog

Fig. 8 Grouping of appliance blocks model under one sub-system.

A photo of the appliance can be used as an icon for the

sub-system block. This gives the user the flexibility of
adapting this system to it appliance without any coding.
Indeed, the user will have to input just the supply frequency
and the file name where the raw current waveform is stored. (b) Initialization
The mask for the sub-system is developed as follows. First, Fig. 9 Sub-system Mask setting
two variables are defined under the “Parameters & Dialog”
The transient current waveform vector is then defined as:
setting tab as shown in Fig. 9 (a). The parameters are the
supply frequency and the current waveform data file name. CurrWavTrans  CurrWav(k1 : k2 ) (4)
The sample number k3 at which the first steady period ends is
In the initialization tab (see Fig. 9(b)), different parameters determined as:
and variables are initialized. The current waveform data
T
‘CurrWav’ and the sampling frequency Fs are first red. Then k3   k 2 (5)
the sampling time waveform period are calculated as, dt
Finally, the steady state current waveform vector is defined as:
1 1
dt  , T  (1) CurrWavSteady  CurrWav(k2 : k3 ) (6)
Fs f Note that in Fig. 9 the data vectors’ names for the transient
Then, the sample k1 at which the switching on took place and steady state waveforms have “_AC” attached at their ends,
is determined by comparing the current to a threshold value. which indicate that the appliance is an air-conditioning unit.
The transient time Ttrans and corresponding sample number k2 Indeed, since the vectors are defined as global and in case
are determined as follows: more than one appliance is modeled in the same system, then
Ttrans  tmax  4T (2) the vectors’ names should be differentiated by an additional
designation, which could be an abbreviation of the name of
Ttrans the appliances. The abbreviations in Table 1 were adopted in
k2  (3)
dt this case.

Authorized licensed use limited to: Auckland University of Technology. Downloaded on May 28,2020 at 08:45:43 UTC from IEEE Xplore. Restrictions apply.
 TABLE 1 ADOPTED ABBREVIATION OF APPLIANCES NAMES B. Simulation Results
Appliance Abbreviation To validate the developed models, several appliances
where simulated simultaneously under one Simulink system
Air Conditioning AC as shown in Fig. 11.
Air Pump AP
Dish Washer DW
Drilling Machine DM
Hair Dryer HD
Cloth Iron CI
Laundry Dryer LD
LED Light LL
Light Bulbs LB
Microwave Oven MW
Mixer MX
Refrigerator FG
Television TV
Vacuum Cleaner VC
Washing Machine WM
Water Heater WH
Water Pump WP

III. SIMULATION RESULTS

A. Current Signature Datasets
One of the few worldwide published and publically
accessible archives of household current waveform
(signature) datasets is the WHITED dataset introduced in [51].
This dataset comprises a broad range of household and small-
industry appliance start-up transients for a line voltage of 230
V and 50 Hz frequency. The waveforms in this dataset were
collected using a computer Sound Card. The signals were
recorded at 44.1 kHz sampling frequency and 16 bit AD
conversion resolution. Fig. 11 Simulation of selected appliances for model validation.

The appliance voltage and current waveforms were

measured for a duration of 5 sec, which includes and covers Air Conditioning unit
the startup transient duration of all experimented appliances.
The proposed measurement system in [51] is considered as a
low-budget and custom measurement system that allows one
to retrieve significantly discriminating features from
appliance start-up transients to enable appliance classification
needs [51].
Washing Machine
A/C current waveform Washing machine current waveform

Microwave Oven

Microwave oven current waveform TV current waveform

TV

Fig. 10 Samples of dataset current waveforms for four selected appliances: Fig. 12 Simulated current waveforms of selected appliances. At 1.2 sec all
Air Conditioning unit, Washing machine, microwave oven and TV. appliances are switched on. At 2.2 s, all appliances are switched off. At 3.2 s
all appliances are switched on again.

Authorized licensed use limited to: Auckland University of Technology. Downloaded on May 28,2020 at 08:45:43 UTC from IEEE Xplore. Restrictions apply.
 [6] S. Barker, A. Mishra, D. Irwin, E. Cecchet, P. Shenoy, and J. Albrecht,
“Smart*: An Open Data Set and Tools for Enabling Research in
A manual switch is implemented to control the switching
Sustainable Homes,” in SustKDD, August 2012.
on and off of the appliances simultaneously. The results [7] M. Zeifman and K. Roth, “Nonintrusive Appliance Load Monitoring:
obtained in Fig. 12 show that each appliance model is able to Review and Outlook,” IEEE Transactions on Consumer Electronics,
generate the exact current waveform during transient and vol. 57, no. 1, February 2011.
[8] A. Molina-Markham, P. Shenoy, K. Fu, E. Cecchet, and D. Irwin,
steady states that correspond exactly to the measured real
“Private Memoirs of a Smart Meter,” in BuildSys, 2010.
current waveforms (see Fig. 10). [9] S. Barker, A. Mishra, D. Irwin, P. Shenoy, and J. Albrecht, “SmartCap:
Flattening Peak Electricity Demand in Smart Homes,” in PerCom,
IV. CONCLUSION March 2012.
[10] J. Lu, T. Sookoor, V. Srinivasan, G. Gao, B. Holben, J. Stankovic, E.
This paper introduced a new method of modeling home Field, and K. Whitehouse, “The Smart Thermostat: Using Occupancy
appliances using Matlab/Simulink while the same approach Sensors to Save Energy in Homes,” in SenSys, November 2010.
can also be used for implementation with other simulation [11] L. R. Rabiner. A tutorial on hidden Markov models and selected
applications in speech recognition. Proceedings of the IEEE, 77(2):257–
tools. The proposed method is essentially based on splitting 286, 1989.
the real current waveform measured during switching on/of [12] G. David Forney Jr. The Viterbi algorithm. Proceedings of the IEEE,
appliances into two parts: transient and steady state. The 61(3):268–278, 1973.
transient part is stored in a data vector and the steady-state data [13] Andrew J. Viterbi. Error bounds for convolutional codes and
asymptotically optimum decoding algorithm. IEEE Transactions on
is stored in another data vector. The Simulink model processes Information Theory, 13(2):260–269, 1967.
the two vectors and an equivalent current waveform is [14] Wang Yue, Tian Yiwei, Ge Jian and Xu Ying, "The study on the status
generated based on the state of operation. The proposed of energy consumption and energy-efficiency management of new rural
residential buildings in Hangzhou, China," 2011 International
modeling approach was tested on several current signatures of
Conference on Business Management and Electronic Information,
several home appliances and the results obtained show that the Guangzhou, 2011, pp. 49-52.
simulated and the actual waveforms are identical. [15] Wang Changfeng, Ren Shengfeng, Zhao Guomin and Fan Guoqiang,
"Energy-saving design strategy of rural residential building in Tianjin
Because the proposed method is using the actual current area," 2011 International Conference on New Technology of
waveforms, it has the advantage of providing much higher Agricultural, Zibo, 2011, pp. 522-527.
accuracy compared to other methods that are based on [16] L. Changliang, W. Yanqun, B. Kang, L. Weiliang and C. Chen, "Energy
management strategy research for residential microgrid considering
mathematical approximation and modeling of the appliances virtual energy storage system at demand side," 2017 13th IEEE
based on their components or current waveforms. The International Conference on Electronic Measurement & Instruments
developed home appliances’ library is very useful for BEMS (ICEMI), Yangzhou, 2017, pp. 273-280.
designers who can simulate and validate their developed [17] M. Jamil and S. Mittal, "Building Energy Management System: A
Review," 2017 14th IEEE India Council International Conference
algorithms using the actual home appliances real dynamic (INDICON), Roorkee, 2017, pp. 1-6.
operation characteristics and performances. [18] E. Bejoy, S. N. Islam and A. M. T. Oo, "Optimal scheduling of
appliances through residential energy management," 2017 Australasian
Even though, the focus is drawn on home appliances, the Universities Power Engineering Conference (AUPEC), Melbourne,
proposed modeling approach can be applied to any electric VIC, 2017, pp. 1-6.
appliance. Most importantly, the actual current waveforms of [19] L. Martirano et al., "Aggregation of Users in a Residential/Commercial
Building Managed by a Building Energy Management System
the appliances are acquired and collected in a library for (BEMS)," in IEEE Transactions on Industry Applications, vol. 55, no.
research purposes. 1, pp. 26-34, Jan.-Feb. 2019.
[20] K. P. N and P. V. D, "Machine Learning Based Residential Energy
V. ACKNOWLEDGMENT Management System," 2017 IEEE International Conference on
Computational Intelligence and Computing Research (ICCIC),
This publication was made possible by NPRP grant Coimbatore, 2017, pp. 1-4.
[NPRP10-1203-160008] from the Qatar National Research [21] Y. Lee, C. Chuang, Y. Chen and C. Hsueh, "An efficient scheduling
mechanism for building energy management systems," 2015 IEEE
Fund (a member of Qatar Foundation) and the co-funding by International Conference on Building Efficiency and Sustainable
IBERDROLA QSTP LLC. The findings achieved herein are Technologies, Singapore, 2015, pp. 17-22.
solely the responsibility of the authors. [22] H. Heydari, S. M. T. Bathaee, A. Fereidunian, E. Heydari and A.
Nazarzadeh, "Energy saving conception of smart grid focusing on air-
conditioning energy management system," 2013 Smart Grid Conference
VI. REFERENCES (SGC), Tehran, 2013, pp. 138-142.
[1] Energy Information Administration. (2019, Jan.). Annual Energy [23] J. M. G. López, E. Pouresmaeil, C. A. Cañizares, K. Bhattacharya, A.
Outlook 2019 with projections to 2050. Office of Energy Analysis, U.S. Mosaddegh and B. V. Solanki, "Smart Residential Load Simulator for
Department of Energy, Washington, DC 20585 [Online]. Available: Energy Management in Smart Grids," in IEEE Transactions on
https://www.eia.gov/outlooks/aeo/pdf/aeo2019.pdf Industrial Electronics, vol. 66, no. 2, pp. 1443-1452, Feb. 2019.
[2] S. Dawson-Haggerty, S. Lanzisera, J. Taneja, R. Brown, and D. Culler, [24] R. Yao, and K. Steemers, “A method of formulating energy load profile
“@scale: Insights from a Large, Long-Lived Appliance Energy WSN,” for domestic buildings in the UK,” Energy and Buildings, vol. 37, pp
in IPSN, April 2012. 663-671, 2005.
[3] T. Hnat, V. Srinivasan, J. Lu, T. Sookoor, R. Dawson, J. Stankovic, and [25] J.V. Paatero, and P.D. Lund, “A model for generating household
K. Whitehouse, “The Hitchhiker’s Guide to Successful Residential electricity load profiles,” Int. J. Energy Research, vol. 30, pp. 273-290,
Sensing Deployments,” in SenSys, November 2011. 2006.
[4] X. Jiang, M. V. Ly, J. Taneja, P. Dutta, and D. Culler, “Experiences with [26] I. Richardson, “A high-resolution domestic building occupancy model
a High-fidelity Building Energy Auditing Network,” in SenSys, for energy demand Simulations,” Energy and Buildings, vol. 40, pp.
November 2009. 1560-1566, 2008.
[5] Y. Kim, T. Schmid, Z. Charbiwala, and M. Srivastava, “ViridiScope: [27] W. Guo and T. Ullah, "Deployment of a load simulator in simulating
Design and Implementation of a Fine Grained Power Monitoring residential household appliances," 2015 IEEE 12th International
System for Homes,” in Ubicomp, September 2009. Conference on Networking, Sensing and Control, Taipei, 2015, pp. 570-
575.

Authorized licensed use limited to: Auckland University of Technology. Downloaded on May 28,2020 at 08:45:43 UTC from IEEE Xplore. Restrictions apply.
[28] A. Prudenzi, A. Silvestri and G. Lucci, "DSM impact prediction through [40] L. Jiang, S. Luo and J. Li, "An Approach of Household Power Appliance
a model of the Italian residential end-use," 2011 16th International Monitoring Based on Machine Learning," 2012 Fifth International
Conference on Intelligent System Applications to Power Systems, Conference on Intelligent Computation Technology and Automation,
Hersonissos, 2011, pp. 1-5. Zhangjiajie, Hunan, 2012, pp. 577-580.
[29] A. Keshtkar and S. Arzanpour, "A fuzzy logic system for demand-side [41] Ning Lu, Yulong Xie, Zhenyu Huang, F. Puyleart and S. Yang, "Load
load management in residential buildings," 2014 IEEE 27th Canadian component database of household appliances and small office
Conference on Electrical and Computer Engineering (CCECE), equipment," 2008 IEEE Power and Energy Society General Meeting -
Toronto, ON, 2014, pp. 1-5. Conversion and Delivery of Electrical Energy in the 21st Century,
[30] H. Hassen, “Implementation of energy hub management system for Pittsburgh, PA, 2008, pp. 1-5.
residential sector,” Master’s thesis, University of Waterloo, April 2010. [42] H. Murata and T. Onoda, "Estimation of power consumption for
[Online]. Available: https://uwspace.uwaterloo.ca/handle/10012/5224 household electric appliances," Proceedings of the 9th International
[31] A. Prudenzi and A. Silvestri, "Simulation of DSM Actions Impact Conference on Neural Information Processing, 2002. ICONIP '02.,
Prediction on Residential Daily Load Shape," 2011 46th International Singapore, 2002, pp. 2299-2303 vol.5.
Universities' Power Engineering Conference (UPEC), Soest, Germany, [43] M. J. Ghorbani, M. S. Rad, H. Mokhtari, M. E. Honarmand and M.
2011, pp. 1-6. Youhannaie, "Residential Loads Modeling by Norton Equivalent Model
[32] A. Sá, R. A. Lopes and J. F. Martins, "Design of an agent-based of Household Loads," 2011 Asia-Pacific Power and Energy Engineering
simulator for real-time estimation of power consumption/generation in Conference, Wuhan, 2011, pp. 1-4.
residential buildings," IECON 2015 - 41st Annual Conference of the [44] C. B. Nilsen, B. Hoff and T. Østrem, "Framework for Modeling and
IEEE Industrial Electronics Society, Yokohama, 2015, pp. 003832- Simulation of Household Appliances," IECON 2018 - 44th Annual
003838. Conference of the IEEE Industrial Electronics Society, Washington,
[33] G. Basso, P. Ferrez, D. Gabioud and P. Roduit, "An Extensible DC, 2018, pp. 3472-3476.
Simulator for Dynamic Control of Residential Area: Case Study on [45] Y. Cheng, B. Qi and X. Lv, "The research of household appliance on-
Heating Control," 2015 Euromicro Conference on Digital System line load monitoring," 2014 China International Conference on
Design, Funchal, 2015, pp. 486-493. Electricity Distribution (CICED), Shenzhen, 2014, pp. 195-199.
[34] A. Keshtkar, S. Arzanpour and F. Keshtkar, "An autonomous system via [46] X. Zhang et al., "IEHouse: A non-intrusive household appliance state
fuzzy logic for residential peak load management in smart grids," 2015 recognition system," 2017 IEEE SmartWorld, Ubiquitous Intelligence
North American Power Symposium (NAPS), Charlotte, NC, 2015, pp. & Computing, Advanced & Trusted Computed, Scalable Computing &
1-6. Communications, Cloud & Big Data Computing, Internet of People and
[35] O. Abrishambaf, M. A. F. Ghazvini, L. Gomes, P. Faria, Z. Vale and J. Smart City Innovation (SmartWorld/SCALCOM/UIC/ATC/CBDCom
M. Corchado, "Application of a Home Energy Management System for /IOP/SCI), San Francisco, CA, 2017, pp. 1-8.
Incentive-Based Demand Response Program Implementation," 2016 [47] T. Liu, X. Ding and N. Gu, "A Generic Energy Disaggregation
27th International Workshop on Database and Expert Systems Approach: What and When Electrical Appliances are Used," 2015 IEEE
Applications (DEXA), Porto, 2016, pp. 153-157. International Conference on Data Mining Workshop (ICDMW),
[36] S. Li and D. Zhang, "Developing smart and real-time demand response Atlantic City, NJ, 2015, pp. 389-397.
mechanism for residential energy consumers," 2014 Clemson [48] P. Du and N. Lu, "Appliance Commitment for Household Load
University Power Systems Conference, Clemson, SC, 2014, pp. 1-5. Scheduling," in IEEE Transactions on Smart Grid, vol. 2, no. 2, pp. 411-
[37] B. Sudiarto, A. N. Widyanto and H. Hirsch, "The effect of mains voltage 419, June 2011.
level variations on the disturbances produced by household appliances [49] V. P. Bhuvana and A. M. Tonello, "Appliance load disaggregation based
in the frequency range of 9–150 kHz," 2017 International Symposium on maximum a-posterior particle filter (MAP-PF)," 2017 IEEE
on Electromagnetic Compatibility - EMC EUROPE, Angers, 2017, pp. International Conference on Smart Grid Communications
1-5. (SmartGridComm), Dresden, 2017, pp. 165-170.
[38] G. Zhang, X. Ji, C. Li, L. Tao and X. Wu, "Research on energy-saving [50] E. M. Malatji and D. Nicolae, "Optimal electricity purchasing model for
control of standby household appliances," 2014 9th IEEE Conference smart homes using genetic algorithm," 2016 IEEE 16th International
on Industrial Electronics and Applications, Hangzhou, 2014, pp. 634- Conference on Environment and Electrical Engineering (EEEIC),
638. Florence, 2016, pp. 1-4.
[39] K. L. Xiang et al., "A residential end-use electric energy consumption [51] Kahl, M., Haq, A. U., Kriechbaumer, T., and Jacobsen, H. A.,
model based on the development of household appliances," 2017 IEEE ‘WHITED- a worldwide household and industry transient energy data
Conference on Energy Internet and Energy System Integration (EI2), set.’ Workshop on Non-Intrusive Load Monitoring (NILM), 2016.
Beijing, 2017, pp. 1-5.

Authorized licensed use limited to: Auckland University of Technology. Downloaded on May 28,2020 at 08:45:43 UTC from IEEE Xplore. Restrictions apply.