Professional Documents
Culture Documents
Deeplearningtechniques
Deeplearningtechniques
Abstract
The field of smart meter data analytics is a relatively young field
that recently grew due to the wealth of data generated from smart
meters. Recent progress in high-performance computing made it possi-
ble to efficiently use deep neural networks to extract more actionable
knowledge and patterns from massive data collected within the smart
grid. Examining the available literature, there is a need for a compre-
hensive survey that covers all aspects of smart meter data analytics.
This review paper provides a broad overview of the current research
spectrum within smart meter analysis, while identifying future chal-
lenges for smart meter data analytic through a detailed taxonomy. The
proposed taxonomy identifies the main domains within the field, as
well as the various functions that are conducted within each domain.
To show the strength of deep learning techniques in smart meter
analytics, and experiment is conducted comparing various classical
load profiling techniques with deep clustering. The results of the
experiment conducted on an open large dataset show the perfor-
mance of deep clustering to be better than classical techniques.
Furthermore, the review is performed from the perspective of data
science; emphasizing the data tasks such as data reprocessing, ana-
lytic algorithms, and evaluation methods for each domain. Further-
more, the paper discusses deep neural network techniques used for
each domain of smart meter analysis and identifies gaps within each
domain, including obstacles and potential opportunities for research.
1
Springer Nature 2021 LATEX template
1 Introduction
The economy of the 21st century grows increasingly dependent on sustainable
use of energy. The smart grid is the first building block for responsible energy
use in the modern society. Power management, forecasting, planning, and many
other functions can only be implemented through the efficient use of the smart
grid and its main component the smart meter.
However, the large volume of raw data generated by smart meters needs to
be stored, cleaned and analysed, to discover commercially usable knowledge
and gain insight. This wealth of information can be utilized by individuals,
the energy industry, service providers, as well as regulators and planners. The
techniques used for data-driven analysis of this data vary. Machine learning
techniques, among other techniques, are widely implemented for this task.
However, high-performance computing made recent advanced techniques such
as deep and reinforcement learning more accessible for many of the data-driven
analysis tasks within the smart grid.
This review paper aims to survey the various deep neural networks architec-
tures and techniques used for smart meter analysis by enumerating the current
research trends as well as provides a better understanding of the available
opportunities for researchers.
The paper first presents an overview of smart meter intelligence plus a
taxonomy of smart meter analytics based on [1] to aid researchers in navigating
the multitude of domains, algorithms, and methods available within the field.
This taxonomy will be the basis of the review of the field of smart meter deep
learning research. Thus, based on a review of the literature, deep learning
techniques used within each domain will be discussed while highlighting the
network architectures selected due to the nature of each task. The last section
of the paper outlines various open research problems as well as challenges and
opportunities available for researchers in the field of deep learning for smart
meter analytics.
AI Amjady &
Methods Keynia 2008 [50]
were categorized and presented in [24]. The following are the main methods
used for load forecasting regardless of horizon and scale:
• Statistical: These techniques can be deterministic models which express the
relationship between the load and several input factors. SARIMAX was used
for load forecasting in [39]
• Time series: This was by far the most popular method for load forecasting
before the machine learning era . However, this usually is limited to univari-
ate time series analysis which is a compromise that may lead to the loss of
important points in the data set [40].
• Artificial intelligence and machine learning: The AI techniques used in load
forecasting include artificial neural networks, deep network, fuzzy systems,
swarm intelligence, and evolutionary computation.
Among Artificial Intelligence methods, deep neural networks are consid-
ered as one of the most promising methods for short-term load forecasting
(STLF) due to two critical characteristics of deep learning: the ability to learn
extremely non-linear interactions and the ability to learn mutual ambiguity
in data sets. This fact is also reflected in Figure 3, which shows that several
papers used deep learning for STLF at the meter level.
The deployment of deep networks at other scales and horizons (other than
STLF) was not addressed in the literature. Deep learning networks architecture
used for STLF include: recurrent neural networks (RNN) [51] and Long Short-
Term Memory (LSTM) [52], Convolutional Neural Networks (CNN) [53], and
Bayesian deep learning [54, 55]. An online RNN model was proposed in [56]
Springer Nature 2021 LATEX template
and an online LSTM model was proposed by [57]. In addition to these, recently
a hybrid network was also was designed for STLF (i.e. LSTM network with
CNN) [58, 59] and an RNN with LSTM [60].
shows DEC [64] raised the Silhouette Index [62] of the clusters from 0.084
for k-Means and 0.464 for Hierarchical Clustering into 0.59 for DEC [64] and
DynAE [65] deep clustering.
The paper [63] outlines a comparison with several other traditional cluster-
ing algorithms and shows that deep clustering outperforms them in obtaining
good load profiles. Please refer to [63] for a detailed comparative study of
the effectiveness of classical versus deep clustering methods for load profiling.
Furthermore, due to the temporal nature of smart meter readings, all tempo-
ral deep clustering networks are potentially suitable for load profiling tasks.
Different temporal deep network were described in [66].
RNN [74],[75]
Certain distance measures are employed to assess the correlations between dif-
ferent pairs of subsequences. Motifs are discovered as sub-sequences with a
significant degree of similarity.
additional advantages from the utility. Physical tampering, such as illegal tap-
ping, bypassing meters, firmware modification, and cyberattacks against the
AMI infrastructure are used to manipulate smart meter reading. Non-technical
loss (NTL) in power systems is mostly the result of electricity theft, billing
problems, and unpaid bills.
various price techniques. Implementing a deep network for this job is a new
research focus, as indicated in Figure 3. Table 3 shows some of the network
architectures used for this task. In addition to pure demand response, Long
Short Term Memory (LSTM) networks were utilized to improve the accuracy
of the forecasting task [95].
data collected and communicated by smart meters may disclose sensitive infor-
mation about households such as habits, lifestyle choices, and even religious
views.
These all emphasize the growing interest in privacy-enhancing technology
in the smart meter analysis. Cryptographic techniques for assuring privacy-
preserving smart meter data management under the trusted operator model, as
well as privacy-preserving solutions for data processing under the non-trusted
operator model was examined in [109]. So, providing data protection along
with smart meter analysis is crucial.
Recently Differential Privacy (DP) algorithms were used to protect per-
sonal information. Differential privacy is a method that allows researchers
and database analysts to extract relevant information from databases holding
people’s personal information without disclosing personal information about
individuals. For example, in [110] DP property added to forecasting imple-
mentation through benefiting from Google’s TensorFlow Privacy library [110].
Table 3 represents some deep networks used for this task.
• CKSC [111]: A hybrid method that uses both k-Means and Spectral
clustering algorithms.
• Dynamic Auto-Encoder (DynAE): A recenyly proposed deep clustering
method for load profiling [63].
Note that while the original set had a total of 6435 houses, in the pre-
processing stage we removed houses with NaN values or zero meter readings
(149 houses). Thus the experiment was conducted on 6286 houses with a time-
series that starts from 1pm on the 14th of July.
The pre-processing phase flipped the data so that the rows are the houses
and the columns represent smart meter readings (for more details please
see [63]. The dimensions of the resulting matrix was 6286 rows and 26010
columns.
The Dynamic Auto-Encoder network used in the experiment was a deep
network with 6286, 2000, 3000, 2000, 1000, 365, 1000, 2000, 3000, 2000, 6286
units respectively. The batch size was 250 and the optimizer used was ADAM.
Table 4 shows the Silhouette Index results for different cluster numbers (k)
and the various algorithms.
Table 4 Silhouette Index (SI) results for classical vs the deep clustering method DynAE
Classical Clustering Deep Clustering
k
k-Means GMM Agglomerative CKSC [111] DynAE
2 0.7829 0.6627 0.5431 0.4250 0.7395
3 0.7040 0.0832 0.1248 0.2330 0.6158
4 0.5032 0.0357 0.0087 0.2400 0.5285
5 0.1744 -0.0277 0.0002 0.1230 0.5090
6 0.1803 -0.0342 -0.1287 0.1230 0.5167
7 0.1728 -0.0463 -0.1345 0.1330 0.5404
8 0.1751 -0.4430 -0.2450 0.1050 0.5204
Note that as explained in [111] and in [63], the SI values for small values of
k are disregarded because these algorithms do not form good clusters for small
values of k and tend to group all customers in one or two clusters. As shown
in Figure 5 and using the elbow method, the best clusters were obtained for
this data set at k = 5 and k = 6. This was also supported by the result of
CKSC investigation [64]. Although k-Means performed well for small values of
k in the experiment, for k = 5 and k = 6 the deep learning DynAE algorithm
performed much better than all the traditional clustering methods (including
the hybrid CKSC algorithm).
As discussed in sub-section 2.2.1, the deep learning algorithm also has the
advantage of better performance with high dimensional data (26010 columns),
and it also performs pre-processing and feature selection automatically [63].
The results in Figure 5 and Table 4 clearly show the strengths of deep learning
for high dimensional data such as smart meter readings for an extended amount
of time.
Springer Nature 2021 LATEX template
4.1 Challenges
The challenges facing deep learning research in the smart-meter field stem
from the nature of big data, as well as the lack of experience and results in
applying deep learning methods within the field.
• Access to Real-World Data: The first key challenge facing researchers in
smart meter data analytics is shared among all data science projects that
require real data sets. Despite the fact that smart meters are widely deployed
in many countries, the quantity of publicly available data sets for research
purposes is quite restricted. The utilities and electric suppliers control the
data and very few made some data sets public, limiting the amount of
relevant real world experiments that can be conducted.
• Variety and Data Granularity: Another data challenge is that each job neces-
sitates a different degree of data granularity, thus access to a different type of
smart meter reading is required for each domain. For example, the load dis-
aggregation task requires data to be read every one second, whereas demand
response requires data to be read every one minute.
Springer Nature 2021 LATEX template
incremental and online learning are two viable options. Since smart meter
data are real-time stream data, methods derived from these techniques, such
as online dictionary learning [80] and incremental machine learning [112],
may be used to analyze continuous streams of smart meter data.
• Edge Computing: As implementing deep learning for smart meter analysis
relies on the instant availability of the required compute resources, smart
meter data collection and storage takes place primarily on the cloud. Net-
work constraints and cloud access may also impede real-time applications.
Edge Computing is a new method to address these drawbacks, which is a
computing model that makes computing and storage resources available at
the network’s edge while simultaneously minimizing latency and bandwidth
consumption; resulting in reduced network delays and congestion [113].
• Deep Clustering and Transfer Learning: While applying deep learning in
smart meter analysis with classification nature shows promising results, it is
an open research field for other tasks such as load profiling. Implementing
state of art networks deployed for other domains, such as computer vision,
natural language, etc; for smart meter analysis is also a new trend in the
field [83]. Besides deep learning, other emerging data science trends, such
as reinforcement learning, privacy preserving transfer learning, and hybrid
machine learning approaches, can be employed in various smart meter anal-
ysis tasks. As shown in Table 1, implementing state of art networks, which
refers to networks that were trained and used for other domain, is a new
approach in the smart meter field.
5 Conclusion
The availability of technologically advanced computer systems and scientific
advances in the field of computing enables the use of complex and computa-
tionally expensive deep learning algorithms to provide successful solutions to a
wide range of engineering problems. Deep neural networks have shown promis-
ing performance in the smart meter analysis field, which is a multidisciplinary
research problem that can benefit from the power of such algorithms.
This review introduced the intersection between smart meter data analyt-
ics and deep neural networks. The body of knowledge was structured using
a new taxonomy that organizes the research into three broad domains: fore-
casting, knowledge discovery, and energy management. The tasks in each area
were explained and reviewed by investigating the deep neural architectures
employed.
The literature survey reveals the growing use of deep learning techniques
in addressing various smart meter analytic application. Due to their flexibility
and scalability, and their excellent classification accuracy, deep neural networks
show promising results for some smart meter analysis tasks as well as emerging
areas such as security and forecasting.
Springer Nature 2021 LATEX template
The review also indicates that, while academics are beginning to apply deep
learning in specific smart meter analysis jobs, building tailored deep networks
for the many domains of smart meter analysis remains an open research field.
In addition to discussing the potential that deep learning networks provide
to smart meter analytics, this paper also demonstrated that other emerging
data science trends, such as reinforcement learning and hybrid machine learn-
ing approaches also can be potentially be used for the task. Implementing a
deep clustering for smart meter tasks is a promising open research area and
has many potential applications.
The comparative experiment presented in Table 4 showed the advantages
of using deep learning for load profiling through better SI results over many of
the well known classical algorithms. Furthermore, deep learning tends to work
well for data with high dimensions and in many cases performs some of the
pre-processing tasks automatically.
The literature also emphasizes the importance of new techniques such as
transfer learning to overcome the limitation of smart meter analysis tasks. This
allows complicated network structures to be trained for other domains in smart
meter analysis. There are many opportunities for new computer technologies
such as edge computing as they are introduced to smart meter analytics tasks,
in order to achieve real-time and online applications of deep learning. In the
future, smart meter analysis may be deployed in edge devices that use deep
learning that has been well-trained on the cloud, as well as other advanced AI
technologies, to accomplish real-time inference.
Declarations
• Acknowledgment: We thank Michael Hutton of the ōbex project for language
editing support.
• Funding: Not applicable
• Conflict of interest: The authors declare that they have no conflict of interest.
References
[1] Eskandarnia, E., Al-Ammal, H.M.: A taxonomy of smart meter analytics:
Forecasting, knowledge discovery, and power management. International
Journal of Computing and Digital Systems, 851–859 (2022)
[2] Goodfellow, I., Bengio, Y., Courville, A.: Deep Learning. MIT press, ???
(2016)
[5] Khuntia, S.R., Tuinema, B.W., Rueda, J.L., van der Meijden, M.A.:
Time-horizons in the planning and operation of transmission networks:
an overview. IET Generation, Transmission & Distribution 10(4), 841–
848 (2016)
[6] Srivastava, A.K., Pandey, A.S., Singh, D.: Short-term load forecast-
ing methods: A review. In: 2016 International Conference on Emerging
Trends in Electrical Electronics & Sustainable Energy Systems (ICE-
TEESES), pp. 130–138 (2016). https://doi.org/10.1109/ICETEESES.
2016.7581373
[7] Sun, L., Zhou, K., Yang, S.: Regional difference of household elec-
tricity consumption: An empirical study of jiangsu, china. Journal of
Cleaner Production 171, 1415–1428 (2018-01-10). https://doi.org/10.
1016/j.jclepro.2017.10.123
[9] Kakran, S., Chanana, S.: Smart operations of smart grids integrated with
distributed generation: A review. Renewable and Sustainable Energy
Reviews 81, 524–535 (2018)
[10] Nafi, N.S., Ahmed, K., Gregory, M.A., Datta, M.: A survey of smart
grid architectures, applications, benefits and standardization. Journal of
Network and Computer Applications 76, 23–36 (2016)
[11] Wang, Y., Chen, Q., Kang, C., Zhang, M., Wang, K., Zhao, Y.: Load
profiling and its application to demand response: A review. Tsinghua
Science and Technology 20(2), 117–129 (2015)
[12] Deb, C., Zhang, F., Yang, J., Lee, S.E., Shah, K.W.: A review on time
series forecasting techniques for building energy consumption. Renewable
and Sustainable Energy Reviews 74, 902–924 (2017-07-01). https://doi.
org/10.1016/j.rser.2017.02.085
[13] Raza, M.Q., Khosravi, A.: A review on artificial intelligence based load
demand forecasting techniques for smart grid and buildings. Renewable
and Sustainable Energy Reviews 50, 1352–1372 (2015-10-01). https://
doi.org/10.1016/j.rser.2015.04.065
[14] Vrablecová, P., Bou Ezzeddine, A., Rozinajová, V., Šárik, S., San-
gaiah, A.K.: Smart grid load forecasting using online support
vector regression. Computers and Electrical Engineering (2018)
Springer Nature 2021 LATEX template
https://arxiv.org/abs/S0045790617320645. https://doi.org/10.1016/j.
compeleceng.2017.07.006
[15] Wei, Y., Zhang, X., Shi, Y., Xia, L., Pan, S., Wu, J., Han, M., Zhao,
X.: A review of data-driven approaches for prediction and classifica-
tion of building energy consumption. Renewable and Sustainable Energy
Reviews 82, 1027–1047 (2018)
[16] Zor, K., Timur, O., Teke, A.: A state-of-the-art review of artificial intel-
ligence techniques for short-term electric load forecasting. In: Energy
(IYCE), 2017 6th International Youth Conference on, pp. 1–7. IEEE,
??? (2017)
[17] Alahakoon, D., Yu, X.: Smart electricity meter data intelligence for
future energy systems: A survey. IEEE Transactions on Industrial
Informatics 12(1), 425–436 (2016)
[18] Yildiz, B., Bilbao, J.I., Dore, J., Sproul, A.B.: Recent advances in the
analysis of residential electricity consumption and applications of smart
meter data. Applied Energy 208, 402–427 (2017-12-15). https://doi.org/
10.1016/j.apenergy.2017.10.014
[19] Liu, X., Golab, L., Golab, W., Ilyas, I.F., Jin, S.: Smart meter data
analytics: systems, algorithms, and benchmarking. ACM Transactions
on Database Systems (TODS) 42(1), 2 (2017)
[20] Yin, X., Li, X., Zhang, Y., Zhang, T., Lu, C., Ai, Q., Li, Z., Sun, Z.: A
survey of deep learning and its application in distribution network. In:
2020 International Conference on Artificial Intelligence in Information
and Communication (ICAIIC), pp. 643–646 (2020). IEEE
[21] Ibrahim, M.S., Dong, W., Yang, Q.: Machine learning driven smart
electric power systems: Current trends and new perspectives. Applied
Energy 272, 115–237 (2020)
[22] Wang, Y., Chen, Q., Hong, T., Kang, C.: Review of smart meter data
analytics: Applications, methodologies, and challenges. IEEE Transac-
tions on Smart Grid (2018)
[23] Petersen, K., Feldt, R., Mujtaba, S., Mattsson, M.: Systematic mapping
studies in software engineering. In: 12th International Conference on
Evaluation and Assessment in Software Engineering (EASE) 12, pp. 1–10
(2008)
[25] Khuntia, S.R., Rueda, J.L., van der Meijden, M.A.: Forecasting the load
of electrical power systems in mid-and long-term horizons: a review. IET
Generation, Transmission & Distribution 10(16), 3971–3977 (2016)
[26] Kong, W., Dong, Z.Y., Hill, D.J., Luo, F., Xu, Y.: Short-term residential
load forecasting based on resident behaviour learning. IEEE Transactions
on Power Systems 33(1), 1087–1088 (2018). https://doi.org/10.1109/
TPWRS.2017.2688178
[27] Kuster, C., Rezgui, Y., Mourshed, M.: Electrical load forecasting models:
A critical systematic review. Sustainable Cities and Society 35, 257–270
(2017-11-01). https://doi.org/10.1016/j.scs.2017.08.009
[28] Yildiz, B., Bilbao, J.I., Sproul, A.B.: A review and analysis of regression
and machine learning models on commercial building electricity load
forecasting. Renewable and Sustainable Energy Reviews 73, 1104–1122
(2017-06-01). https://doi.org/10.1016/j.rser.2017.02.023
[29] Dedinec, A., Filiposka, S., Dedinec, A., Kocarev, L.: Deep belief net-
work based electricity load forecasting: An analysis of macedonian case.
Energy 115, 1688–1700 (2016)
[30] Marino, D.L., Amarasinghe, K., Manic, M.: Building energy load fore-
casting using deep neural networks. In: Industrial Electronics Society,
IECON 2016-42nd Annual Conference of the IEEE, pp. 7046–7051.
IEEE, ??? (2016)
[31] Ma, W., Fang, S., Liu, G., Zhou, R.: Modeling of district load forecasting
for distributed energy system. Applied Energy 204, 181–205 (2017-10-
15). https://doi.org/10.1016/j.apenergy.2017.07.009
[32] Kim, M., Park, S., Lee, J., Joo, Y., Choi, J.K.: Learning-based adaptive
imputation method with kNN algorithm for missing power data. Energies
10(10) (2017-10). https://doi.org/10.3390/en10101668
[33] Li, S., Wang, P., Goel, L.: Short-term load forecasting by wavelet trans-
form and evolutionary extreme learning machine. Electric Power Systems
Research 122, 96–103 (2015)
[34] Dang-Ha, T.-H., Bianchi, F.M., Olsson, R.: Local short term electricity
load forecasting: Automatic approaches. In: Neural Networks (IJCNN),
2017 International Joint Conference on, pp. 4267–4274. IEEE, ??? (2017)
[37] Hu, Z., Bao, Y., Chiong, R., Xiong, T.: Mid-term interval load forecasting
using multi-output support vector regression with a memetic algorithm
for feature selection. Energy 84, 419–431 (2015-05-01). https://doi.org/
10.1016/j.energy.2015.03.054
[38] Koprinska, I., Rana, M., Agelidis, V.G.: Correlation and instance based
feature selection for electricity load forecasting. Knowledge-Based Sys-
tems 82, 29–40 (2015-07-07) https://arxiv.org/abs/S0950705115000714.
https://doi.org/10.1016/j.knosys.2015.02.017
[40] Eskandar, E.M., Alammal, H., Almadany, W.: What prophet says about
electrical consumption-forecasting techniques for big temporal data
(2021)
[41] Liu, C., Jin, Z., Gu, J., Qiu, C.: Short-term load forecasting using a long
short-term memory network. In: 2017 IEEE PES Innovative Smart Grid
Technologies Conference Europe (ISGT-Europe), pp. 1–6 (2017-09-26).
https://doi.org/10.1109/ISGTEurope.2017.8260110
[43] Chen, T.: A collaborative fuzzy-neural approach for long-term load fore-
casting in taiwan. Computers & Industrial Engineering 63(3), 663–670
(2012)
[45] Lee, W.-J., Hong, J.: A hybrid dynamic and fuzzy time series model
for mid-term power load forecasting. International Journal of Electrical
Power & Energy Systems 64, 1057–1062 (2015-01-01). https://doi.org/
10.1016/j.ijepes.2014.08.006
[47] Ghofrani, M., Hassanzadeh, M., Etezadi-Amoli, M., Fadali, M.S.: Smart
meter based short-term load forecasting for residential customers. In:
North American Power Symposium (NAPS), 2011, pp. 1–5. IEEE, ???
(2011)
[48] Hong, T., Wilson, J., Xie, J.: Long term probabilistic load forecasting
and normalization with hourly information. IEEE Transactions on Smart
Grid 5(1), 456–462 (2014). https://doi.org/10.1109/TSG.2013.2274373
[49] Lusis, P., Khalilpour, K.R., Andrew, L., Liebman, A.: Short-term
residential load forecasting: Impact of calendar effects and forecast gran-
ularity. Applied Energy 205, 654–669 (2017-11). https://doi.org/10.
1016/j.apenergy.2017.07.114
[50] Amjady, N., Keynia, F.: Mid-term load forecasting of power systems
by a new prediction method. Energy Conversion and Management
49(10), 2678–2687 (2008-10-01). https://doi.org/10.1016/j.enconman.
2008.04.008
[51] Yuan, L., Ma, J., Gu, J., Wen, H., Jin, Z.: Featuring periodic correlations
via dual granularity inputs structured rnns ensemble load forecaster.
International Transactions on Electrical Energy Systems 30(11), 12571
(2020)
[52] Yang, Y., Hong, W., Li, S.: Deep ensemble learning based probabilistic
load forecasting in smart grids. Energy 189, 116–324 (2019)
[53] Fahiman, F., Erfani, S.M., Rajasegarar, S., Palaniswami, M., Leckie,
C.: Improving load forecasting based on deep learning and k-shape
clustering. In: 2017 International Joint Conference on Neural Networks
(IJCNN), pp. 4134–4141 (2017). IEEE
[54] Yang, Y., Li, W., Gulliver, T.A., Li, S.: Bayesian deep learning-based
probabilistic load forecasting in smart grids. IEEE Transactions on
Industrial Informatics 16(7), 4703–4713 (2019)
[55] Sun, M., Zhang, T., Wang, Y., Strbac, G., Kang, C.: Using bayesian
deep learning to capture uncertainty for residential net load forecasting.
IEEE Transactions on Power Systems 35(1), 188–201 (2019)
[56] Fekri, M.N., Patel, H., Grolinger, K., Sharma, V.: Deep learning for
load forecasting with smart meter data: Online adaptive recurrent neural
network. Applied Energy 282, 116–177 (2021)
[57] Razavi, S.E., Arefi, A., Ledwich, G., Nourbakhsh, G., Smith, D.,
Minakshi, M.: From load to net energy forecasting: Short-term residen-
tial forecasting for the blend of load and pv behind the meter. IEEE
Springer Nature 2021 LATEX template
Access (2020)
[58] Madhure, R.U., Raman, R., Singh, S.K.: Cnn-lstm based electricity
theft detector in advanced metering infrastructure. In: 2020 11th Inter-
national Conference on Computing, Communication and Networking
Technologies (ICCCNT), pp. 1–6 (2020). IEEE
[59] Khan, Z.A., Hussain, T., Ullah, A., Rho, S., Lee, M., Baik, S.W.: Towards
efficient electricity forecasting in residential and commercial buildings:
A novel hybrid cnn with a lstm-ae based framework. Sensors 20(5), 1399
(2020)
[60] Bouktif, S., Fiaz, A., Ouni, A., Serhani, M.A.: Multi-sequence lstm-rnn
deep learning and metaheuristics for electric load forecasting. Energies
13(2), 391 (2020)
[61] Aghabozorgi, S., Seyed Shirkhorshidi, A., Ying Wah, T.: Time-series
clustering – a decade review. Information Systems 53, 16–38 (2015-10-
01). https://doi.org/10.1016/j.is.2015.04.007
[62] Arbelaitz, O., Gurrutxaga, I., Muguerza, J., Pérez, J.M., Perona, I.:
An extensive comparative study of cluster validity indices. Pattern
recognition 46(1), 243–256 (2013)
[64] Xie, J., Girshick, R., Farhadi, A.: Unsupervised deep embedding for clus-
tering analysis. In: International Conference on Machine Learning, pp.
478–487 (2016). PMLR
[65] Mrabah, N., Bouguessa, M., Ksantini, R.: Adversarial deep embedded
clustering: on a better trade-off between feature randomness and feature
drift. IEEE Transactions on Knowledge and Data Engineering (2020)
[66] Aljalbout, E., Golkov, V., Siddiqui, Y., Strobel, M., Cremers, D.: Clus-
tering with deep learning: Taxonomy and new methods. arXiv preprint
arXiv:1801.07648 (2018)
[67] Feng, C., Mehmani, A., Zhang, J.: Deep learning-based real-time build-
ing occupancy detection using ami data. IEEE Transactions on Smart
Grid 11(5), 4490–4501 (2020)
[69] Paresh, S., Thokala, N.K., Chandra, M.G.: Electrical load disaggre-
gation using a two-stage deep learning approach. In: Proceedings of
the 6th ACM International Conference on Systems for Energy-Efficient
Buildings, Cities, and Transportation, pp. 366–367 (2019)
[70] He, W., Chai, Y.: An empirical study on energy disaggregation via deep
learning. In: 2016 2nd International Conference on Artificial Intelligence
and Industrial Engineering (AIIE 2016), pp. 338–342 (2016). Atlantis
Press
[71] Ullah, A., Haydarov, K., Ul Haq, I., Muhammad, K., Rho, S., Lee,
M., Baik, S.W.: Deep learning assisted buildings energy consumption
profiling using smart meter data. Sensors 20(3), 873 (2020)
[72] Zhang, P., Cheng, H., Zou, B., Dai, P., Ye, C.: Load data mining
based on deep learning method. In: Proceedings of the 3rd International
Conference on Computer Science and Application Engineering, pp. 1–5
(2019)
[73] Gong, X., Tang, B., Zhu, R., Liao, W., Song, L.: Data augmentation
for electricity theft detection using conditional variational auto-encoder.
Energies 13(17), 4291 (2020)
[74] Kelly, J., Knottenbelt, W.: Neural nilm: Deep neural networks applied
to energy disaggregation. In: Proceedings of the 2nd ACM Interna-
tional Conference on Embedded Systems for Energy-efficient Built
Environments, pp. 55–64 (2015)
[75] Nabil, M., Ismail, M., Mahmoud, M., Shahin, M., Qaraqe, K., Serpedin,
E.: Deep learning-based detection of electricity theft cyber-attacks in
smart grid ami networks. In: Deep Learning Applications for Cyber
Security, pp. 73–102. Springer, ??? (2019)
[76] Wang, Y., Chen, Q., Gan, D., Yang, J., Kirschen, D.S., Kang, C.: Deep
learning-based socio-demographic information identification from smart
meter data. IEEE Transactions on Smart Grid (2018)
[78] Li, J., Wang, F.: Non-technical loss detection in power grids with sta-
tistical profile images based on semi-supervised learning. Sensors 20(1),
236 (2020)
[79] Hassani, H., Farajzadeh-Zanjani, M., Razavi-Far, R., Saif, M., Palade,
Springer Nature 2021 LATEX template
[80] Singh, S., Majumdar, A.: Deep sparse coding for non–intrusive load
monitoring. IEEE Transactions on Smart Grid 9(5), 4669–4678 (2017)
[81] Devlin, M.A., Hayes, B.P.: Load identification and classification of activ-
ities of daily living using residential smart meter data. In: 2019 IEEE
Milan PowerTech, pp. 1–6 (2019). IEEE
[82] Devlin, M.A., Hayes, B.P.: Non-intrusive load monitoring and classifica-
tion of activities of daily living using residential smart meter data. IEEE
Transactions on Consumer Electronics 65(3), 339–348 (2019)
[83] Farkas, M., Lacko, P.: Using advanced audio generating techniques to
model electrical energy load. In: International Conference on Engineering
Applications of Neural Networks, pp. 39–48 (2017). Springer
[84] Singhal, V., Maggu, J., Majumdar, A.: Simultaneous detection of multi-
ple appliances from smart-meter measurements via multi-label consistent
deep dictionary learning and deep transform learning. IEEE Transactions
on Smart Grid 10(3), 2969–2978 (2018)
[86] Mahmud, A.A., Sant, P.: Real-time price savings through price sugges-
tions for the smart grid demand response model. In: Smart Grid and
Cities Congress and Fair (ICSG), 2017 5th International Istanbul, pp.
65–69. IEEE, ??? (2017)
[87] Längkvist, M., Karlsson, L., Loutfi, A.: A review of unsupervised feature
learning and deep learning for time-series modeling. Pattern Recognition
Letters 42, 11–24 (2014)
[88] Hosseini, S.S., Agbossou, K., Kelouwani, S., Cardenas, A.: Non-intrusive
load monitoring through home energy management systems: A com-
prehensive review. Renewable and Sustainable Energy Reviews 79,
1266–1274 (2017-11-01). https://doi.org/10.1016/j.rser.2017.05.096
[90] Basu, K., Debusschere, V., Bacha, S., Maulik, U., Bondyopadhyay,
S.: Nonintrusive load monitoring: A temporal multilabel classification
approach. IEEE Transactions on industrial informatics 11(1), 262–270
(2015)
[94] Vercamer, D., Steurtewagen, B., Van den Poel, D., Vermeulen, F.: Pre-
dicting consumer load profiles using commercial and open data. IEEE
Transactions on Power Systems 31(5), 3693–3701 (2016)
[95] Christensen, M.H., Nozal, D.C., Kavadakis, I., Pinson, P.: Data-driven
learning from dynamic pricing data-classification and forecasting. In:
2019 IEEE Milan PowerTech, pp. 1–6 (2019). IEEE
[96] Rodriguez Fernandez, M., Garcia, A.C., Alonso, I.G., Casanova, E.Z.:
Using the big data generated by the smart home to improve energy
efficiency management. Energy Efficiency 9(1), 249–260 (2016-01). https:
//doi.org/10.1007/s12053-015-9361-3
[97] Ye, Y., Qiu, D., Wu, X., Strbac, G., Ward, J.: Model-free real-time
autonomous control for a residential multi-energy system using deep rein-
forcement learning. IEEE Transactions on Smart Grid 11(4), 3068–3082
(2020)
[98] Shateri, M., Messina, F., Piantanida, P., Labeau, F.: Privacy-cost man-
agement in smart meters using deep reinforcement learning. In: 2020
IEEE PES Innovative Smart Grid Technologies Europe (ISGT-Europe),
pp. 929–933 (2020). IEEE
[99] Patyn, C., Reymond, M., Rădulescu, R., Deconinck, G., Nowé, A.:
Reinforcement learning for demand response of domestic household
appliances. Adaptive Learning Agents 2018 Proceedings, 1–7 (2018)
[100] Lu, R., Hong, S.H., Zhang, X.: A dynamic pricing demand response algo-
rithm for smart grid: reinforcement learning approach. Applied Energy
220, 220–230 (2018)
Springer Nature 2021 LATEX template
[101] Zhou, X., Canady, R., Li, Y., Koutsoukos, X., Gokhale, A.: Overcom-
ing stealthy adversarial attacks on power grid load predictions through
dynamic data repair. In: International Conference on Dynamic Data
Driven Application Systems, pp. 102–109 (2020). Springer
[103] Marulli, F., Visaggio, C.A.: Adversarial deep learning for energy man-
agement in buildings. In: SummerSim, pp. 50–1 (2019)
[104] Zhou, L., Ouyang, X., Ying, H., Han, L., Cheng, Y., Zhang, T.:
Cyber-attack classification in smart grid via deep neural network. In:
Proceedings of the 2nd International Conference on Computer Science
and Application Engineering, pp. 1–5 (2018)
[105] Kong, X., Kong, D., Yao, J., Bai, L., Xiao, J.: Online pricing of demand
response based on long short-term memory and reinforcement learning.
Applied Energy 271, 114945 (2020)
[106] Liu, M., Liu, D., Sun, G., Zhao, Y., Wang, D., Liu, F., Fang, X., He, Q.,
Xu, D.: Deep learning detection of inaccurate smart electricity meters: A
case study. IEEE Industrial Electronics Magazine 14(4), 79–90 (2020)
[107] Niu, X., Li, J., Sun, J., Tomsovic, K.: Dynamic detection of false data
injection attack in smart grid using deep learning. In: 2019 IEEE Power &
Energy Society Innovative Smart Grid Technologies Conference (ISGT),
pp. 1–6 (2019). IEEE
[108] Jindal, A., Aujla, G.S., Kumar, N., Prodan, R., Obaidat, M.S.: Drums:
Demand response management in a smart city using deep learning and
svr. In: 2018 IEEE Global Communications Conference (GLOBECOM),
pp. 1–6 (2018). IEEE
[109] Asghar, M.R., Dán, G., Miorandi, D., Chlamtac, I.: Smart meter data
privacy: A survey. IEEE Communications Surveys & Tutorials 19(4),
2820–2835 (2017)
[110] Soykan, E.U., Bilgin, Z., Ersoy, M.A., Tomur, E.: Differentially private
deep learning for load forecasting on smart grid. In: 2019 IEEE Globecom
Workshops (GC Wkshps), pp. 1–6 (2019). IEEE
[111] Sun, L., Zhou, K., Yang, S.: An ensemble clustering based framework
for household load profiling and driven factors identification. Sustainable
Springer Nature 2021 LATEX template
[112] Sehovac, L., Nesen, C., Grolinger, K.: Forecasting building energy con-
sumption with deep learning: A sequence to sequence approach. In: 2019
IEEE International Congress on Internet of Things (ICIOT), pp. 108–116
(2019). IEEE
[113] Taı̈k, A., Cherkaoui, S.: Electrical load forecasting using edge com-
puting and federated learning. In: ICC 2020-2020 IEEE International
Conference on Communications (ICC), pp. 1–6 (2020). IEEE