Survey on Technologies Driving the Smart Energy Sector

Full text

29th Telecommunications forum TELFOR 2021 Serbia, Belgrade, November 23-24, 2021.
978-0-7381-4243-2/20/$31.00 ©2020 IEEE
 Abstract — The European electricity system undergoes
significant changes driven by the EU common rules for the
internal market for electricity, as well as by the climate action
agenda. The future smart electricity system will build upon a
combination of a broad range of new and old technologies
together working in novel ways.
Keywords — Smart Grid, Energy System Integration,
Energy Efficiency, Big Data, Artificial Intelligence.
I. INTRODUCTION
he Smart Energy Management refers to a variety of
novel concepts and technologies, serving at both
energy generation and consumption side, such as energy
efficiency, demand management, Smart Grid, micro-grids,
renewable energy sources, and other emerging solutions. It
represents one of the fastest developing fields, according to
the EU priorities, while, at the same time, it remains
somewhat neglected in the South-eastern Europe countries.
Due to the changing framework conditions, like the
liberalization of the energy markets and new regulatory
rules for boosting competitiveness through interoperability
and standardisation, novel approaches for design, planning,
and operation of the electric energy system are under
development.
The goal of this paper is to provide brief insights in the
current state-of-the art fields Smart Grids Landscape and
Energy Efficient Building Operation Landscape in light of
the recently proposed EU Strategy for Energy System
Integration [1]. The research is motivated by the need to
advance the Institute Mihajlo Pupin proprietary VIEW4
SCADA that is currently deployed at many parts of the
Serbian electricity value chain starting from control on
production side (in the large hydro and thermal power
systems), via transmission management to distribution and
electricity dispatching. Hence, the technology enablers
(Big Data, Artificial Intelligence, Edge computing,
Hardware in the loop), that support digitalization and
accelerate the implementation of digital solutions for
building the Common European energy data space,
announced in the European Data strategy [2], are also
The research presented in this paper is partly financed by the
European Union (H2020 SINERGY, Pr. No: 952140), and partly by the
Ministry of Science and Technological Development of the Republic of
Serbia and Science Fund of Republic of Serbia (Artemis). Valentina
Janev, Institute Mihajlo Pupin, University of Belgrade, Volgina 15,
11060 Belgrade, Serbia (e mail: valentina.janev@pupin.rs).
included in the analysis.
In the next Sections, we will first point to the policy
context in Europe and Serbia and the challenges foreseen
until 2030 (Section 2) and then survey the landscape of
technologies from two perspectives: Smart Grids (Section
3) and Energy Efficient Building Operation (Section 4).
Additionally, in Section 5 we discuss the general
Information and Communication Technologies (ICT).
II. BARRIERS FOR ENERGY SYSTEM INTEGRATION
A first priority in the European Union (EU) Political
Agenda for the next period (2019-2024) is the European
Green Deal strategy (2019) [3] that aims to position
Europe as first climate-neutral continent by year 2050. The
energy system is crucial to deliver on the European Green
Deal goals, hence a new document – the EU Strategy for
Energy System Integration [1] was adopted in July 2020
that envisions a coordinated planning and operation of the
energy system ‘as a whole’, across multiple energy
carriers, infrastructures, and consumption sectors. The EU
action plan for transition towards a more integrated energy
system build upon six pillars that address existing barriers
for energy system integration as follows
 A more circular energy system, with ‘energy-efficiency-
first at its core
 Accelerating the electrification of energy demand,
building on a largely renewables-based power system
 Promote renewable and low-carbon fuels, including
hydrogen, for hard-to-decarbonise sectors
 Making energy markets fit for decarbonisation and
distributed resources
 A more integrated energy infrastructure
 A digitalized energy system and a supportive innovation
framework
Unlike previous decades when electricity was mainly
supplied from conventional energy sources such as fossil
fuels (coal, gas) and nuclear energy, renewable energy
sources (RES) in combination with battery storage systems
will be predominant in future smart grids. Reaching the
goal of a high share of electricity generated by variable
RES in Europe - whilst maintaining the high level of
quality and security of supply - will result in an increased
need of flexibility provided by different sources to be
considered in power system planning and operation. Hence
in the next two sections we will refer to the challenges
from two perspectives, the smart grid operation and
maintenance and Energy Efficient Building Operation.
Survey on Technologies
Driving the Smart Energy Sector
Johannes Stöckl, Markus Makoschitz, Thomas Strasser, Luis M. Blanes,
Valentina Janev, Paulo Lissa, Federico Seri
T

III. SMART GRIDS LANDSCAPE
A. Power Electronics and System Components
To establish the Smart Grid, power electronics
converters are seen as key enablers to couple all kinds of
generation and loads effectively. This is driven by the main
technologies for renewable energy generation, which are
either direct current based (e.g. PV) or working with
variable frequency (e.g. Wind). While the first inverters
were mainly power processing units, it became obvious
that the increasing amount of units connected to the grid,
started to influence the grid operation. Due to the volatile
nature of wind or PV-based energy generation, also storage
systems attract more and more attention these days. With
an increasing share of renewables in national grids
(+27 TWh in Austria until 2030) energy storage slowly but
surely becomes an integral part of modern grids as well,
required to flatten and absorb short time energy generation
peaks.In addition, system operators started to establish grid
connection rules to stabilize the system in case of
malfunctions. Today, a large set of rules such as droop
controls, Q(U), Low-Voltage-Ride-Through (LVRT) etc.
have been established worldwide with notable national
differences.
The power electronics design usually depends on many
different critical parameters such as for example
efficiency, dynamic, cost effectiveness, power density,
input/output voltage or input/output current capability. The
degree of relevance assigned to these individual
parameters usually depends on application, location and
infrastructure requirements. A fuel cell, for example, often
requires less than 100 V and several hundred amps input
current, while string inverters normally need 700 V to
1000 V for an input. Cell-phones on the other hand
demand for 5 – 12 V input voltage whereas most Laptop
chargers supply 20 V to the connected device. If fuel cells
are implemented in a car, power density is more relevant
than efficiency. PV inverters on the other hand are often
tuned already for high efficiency (97 - 98 %). For cell-
phone and laptop chargers, both power density and
efficiency is key and becomes more and more important as
they are expected to be easily portable and still should not
generate excessive heat during operation. Furthermore,
with new semiconductor materials such as Silicon Carbide
(SiC) or Gallium Nitride (GaN) on the rise, specific power
electronics applications are in an ongoing transformation
process. Therefore, more and more products on the market
based on this technology can be expected in the future
pushing the existing physical limits adhering to Silicon
based solutions even further. An application readiness map
for commercially available GaN transistors 100 V-650 V is
depicted in Fig. 1. As can be seen, the adoption is affecting
different sectors from grid connected applications,
transportation and even the robotics sector.
These developments towards GaN and SiC materials
also attracting new components with extremely high
voltage ratings (10 kV SiC MOSFETs), combined with the
ideas of smart grids planning and operation, evolves new
system architectures, where low voltage and medium
voltage DC grids received increasing attention during the
last years. Here DC/DC converters as well as suitable
safety components became an additional research topic.
Due to the high efficiency potential new topologies such as
dual active bridges and resonant LLC converters are
investigated, including their potential of grid support on
DC level.
Fig. 1. GaN-based Application Readiness Map (ARM) as published in [4], highlighting relevant grid connected
applications showing significant and predominant market shares until 2035.

B. Power System Digitalization and Automation
The smart grid approach usually leads to an increased
level of complexity of system operation and management.
This also requires distributed intelligence on different
levels in the system. The different levels of intelligence
applied to smart grid systems can be categorized as follows
(see Fig. 2) [5]:
 System level: Approaches like power utility automation,
demand-side management or energy management are
tackled by this level. Functions of the underlying sub-
systems and components are triggered in a coordinated
manner for execution from a systems perspective. Both,
central as well as distributed control approaches are
used on this level.
 Sub-system level: The optimization and the control of
sub-systems are carried out below the system level
whereas the corresponding functions, services, and
algorithms have to deal with a limited amount of
components (RES, energy storage system, electric
vehicle supply equipment, etc.). Examples for this level
are micro-grid control approaches and home/building
energy management concepts. Also an energy storage
system together with a distributed generator installed at
the customer side can be considered as a sub-system.
Distributed automation and control are typically
applied.
 Component level: Distributed Energy Resources
(DER)/RES, distributed energy storage systems but
also electric vehicle supply equipment is covered by
this layer. Components typically provide advanced
functions like ancillary services. Intelligence on this
level is either used for local optimization purposes
(device/component behaviour) or for the optimization
of systems/sub-systems on higher levels in a
coordinated manner.
 Sub-component level: Intelligence on this level is
mainly used to improve the local component
behaviour/properties (harmonics, flicker, etc.). Power
electronics (and their advanced control algorithms) are
the main driver for local intelligence on this level. The
controllers of DER, distributed energy storage systems,
electric vehicle supply equipment and other power
system components (tap-changing transformers,
FACTS, etc.) can be considered as examples for sub-
components.
The rise of the smart grid brought many new
applications (distributed automation / generation / storage,
automated demand response, etc.) that are distributed by
nature and which operate on top of communication
infrastructures. To address the growing demand for
communication due to the various new smart grid
applications as well as for economic reasons, utilities need
to move away from purpose-built networks and consolidate
communication in an integrated smart grid communication
architecture. This transformation is currently ongoing.
Although the utility industry is still (partly) sceptic towards
packet-switching technology (partly due to the “softer”
stochastic QoS guarantees), the IP protocol will be the
overall network protocol [6].
Fig. 2. Necessary intelligence in a smart grid system on
different levels; from [5]
C. Future Challenges
In the area of Smart Grids future challenges are seen on
all system levels. Especially, as the future grid will not be
decoupled from other energy systems but merely play an
important role in an integrated energy system with a further
increasing share in covering the society’s energy needs in
order to fulfil the EU CO2 targets.
With focus on the electric power system the further
increasing penetration with power electronics based
generation and loads leads to a lack of inertia to stabilize
the grid. New control schemes are therefore needed to
ensure power quality. Another approach is the use of DC
technologies as a basis for a new electric grid which can
either be based on pure DC technology or be embedded
synergistically in a hybrid ACDC infrastructure. In any
case, components as well as an efficient grid topology
including grounding schemes are about to be developed.
Operation and control of those infrastructures can be
performed using digital automation technologies.
Parameters (voltages, currents, etc.) are measured,
processed and actions defined based on control strategies
implemented via control algorithms. Typical challenges are
based on big data methodologies to process measurement
data, generation of estimated values for grid nodes (state
estimation) and artificial intelligence methods in place of
classic algorithms to control the grid. The goal is to
provide means for fast response to grid disturbances (e.g.
congestion), enabling the establishment of local energy
communities including local markets.
As most renewable generation technologies need to be
connected via power electronics converters, these
components need to be designed to be grid supportive,
efficient, cost effective, smart and flexible. One important
lever for efficiency is seen in the mass roll-out of wide
band-gap semiconductors (SiC/GaN). Artificial
intelligence methods will also be a backbone for future
developments on various levels [15]. Optimization is

needed for design processes (e.g. smart routing of PCBs,
heat sink optimization), controls (parameter optimization,
self-learning control algorithms) but also big data
methodologies will be needed to provide cost effective
predictive maintenance.
IV. ENERGY EFFICIENT BUILDING OPERATION LANDSCAPE
There are three approaches to more efficient building
energy efficiency [7]: (1) Monitoring the operational stage
of building lifecycle (BLC), (2) Using smart tools and
methods for enhanced building operation; and (3)
Analyzing the users’ behaviour, patterns and implementing
technologies to improve their perception of the building
and provide with a more details on how savings could be
achieved.
The BLC approach considers building impacts in energy
consumption but also the use of energy and environmental
impacts needed to produce the construction materials
(embodied energy), to construct the building itself, and the
need for a demolition and recycling stage to be considered.
Currently, there is a shift towards a circular view of this
lifecycle, which represents a major paradigm change,
including a reuse and zero-waste approaches, instead of the
current linear model.
There are several approaches for improving the
operational energy used in buildings such as using
 Fault Detection, Diagnosis and Prognosis of HVAC
systems;
 Model Predictive Control (MPC) algorithms for optimal
control of various parameters (temperatures, valve
positions, flow rates, pressures, etc.) taking into
account a dynamic model, its utilization for enhanced
building operation;
 AI services which supplement each other (e.g. edge
services, cloud AI-enabled services for predictive
maintenance, semantic technologies, advanced
optimization algorithms);
Finally, the behavioural-based buildings operational
efficiency techniques seek to design the right incentive
based policies that enable and take advantage of the new
paradigms of the energy transition. The human element is a
key factor especially when considering the new scenario
that brings the liberalization of the retail energy markets
with retailers competing to provide enhanced products, the
improvements on domestic scale generation and storage.
Future challenges in building management systems are
related to Big data management and implementation of
behavioural based efficiency measures. Regarding
addressing behavioural based efficiency measures, the
current state of the art [16] highlights the importance of
advancing in the construction and standardization of
typical energy behavioural models by using realistic
consumption data, the use of big data techniques to
improve the understanding of real energy consumption
behaviour and the enhancement of methods to quantify the
energy savings due to the effects of behaviour changes. A
response to these challenges is currently undertaking a
thorough revision by the international initiative IEA EBC
ANNEX 66 “Definition and simulation of occupant
behaviour in buildings” and its follow-up IEA EBC Annex
79 “Occupant-centric building design and operation.” [17].
This project recognises the dynamic and interdependent
nature of interactions between buildings and their
performance and the need to build new models of occupant
centric control. Interestingly, one of the challenges
highlighted in these ANNEX preliminary works and other
reviews [18] is the need for an enhanced user interface that
adequately bridges the knowledge gap between occupants,
operators and KPIs in place and offers timely visualization
of consumption trends, performance, faults and occupant
feedback.
V. EMERGING ICT TECHNOLOGIES AND TRENDS
The transition to a more integrated energy system is of
crucial importance for Europe. With solar and wind power
on the rise, grid operators need new equipment to make the
whole power system operate flexibly [9]. Hence novel
sensors, advanced data exchange infrastructures, and data
handling capabilities that make use of Big Data, Artificial
Intelligence, 5G and distributed ledger technologies are
needed to enhance forecasting, allow the remote
monitoring and management of distributed generation and
improve asset optimisation. Herein, we define these
emerging technological trends and their contribution to
smart energy sector.
A. Interoperability, Standardization, Data Spaces
Data in the energy domain grows at unprecedented rates.
Despite the great potential that IoT platforms and other big
data-driven technologies have brought in the energy sector,
data exchange and data integration are still not wholly
achieved. As a result, fragmented applications are
developed against energy data silos, and data exchange is
limited to few applications.
EU foresees transformation of existing industry value
chain processes, introduction of intermediaries and
marketplaces, introduction of SOA platforms that address
both dimensions of change
 integrated business processes supported by cross-
functional teams; and
 integrated tools along the entire value chain that enable
the processes.
Several initiatives have been proposed to develop
effective and scalable semantic interoperability towards
data spaces. In the context of the European Data Strategy
[2] and the proposed Regulation on European Data
Governance [8], a vision has been created for trusted data
intermediaries for B2B data sharing and common
European data spaces, as well as for promoting advanced
energy services via a marketplace (see for instance EU
projects FLEXICIENCY ,
https://ec.europa.eu/inea/en/horizon-2020/projects/h2020-
energy/grids/flexiciency, BD4OPEM, https://bd4opem.eu/
and PLATOON, https://platoon-project.eu/. Adopting
semantic interoperability solutions, various interconnected
systems and data sources will be described using a
common representation language, thus achieving an

unified access to the available data.
B. Big Data and IoT
Data in the energy domain (even greater now with the
introduction of smart grids) are heterogeneous, with
varying resolution, mostly asynchronous, and are stored in
different formats (raw or processed) at various locations.
For example, typical smart meter data are energy demand
collected every 15 minutes and are stored in billing
centres. One million smart meters installed in utility results
to nearly 3 TB of new energy demand data every year [10].
When faced with thousands, if not millions, of
measurement and control points as it would be the case of
power grids (down to Low voltage), current centralized
data processing approach starts losing efficiency and
facing serious limitations:
 More than 80% of the information sent for central
processing has little, if any value, but cannot be directly
discarded until processed.
 Communications are used to their limit, and available
bandwidth consumed with data that may not be
valuable.
 Cost for central processing and data processing
increases exponentially.
 Reaction times to all this massive influx of data is
continually increasing, thus limiting the value of Big
data.
The continuous lowering of the cost of industrial
computing equipment and the deployment of new
distributed IoT/Edge architectures provides, in a unique
manner, answers to the Big Data Challenge posed to Power
companies. Hence, in response to the challenge of turning
large volumes of raw data into actionable information new
software architectures have been proposed that move from
central data processing and technologies to distributed
processing, edge computing and application of artificial
intelligence techniques close to the source of data.
C. Artificial Intelligence in Smart Grids Applications
A number of Smart Grid applications currently cannot
be implemented without a grid monitoring based on:
 advanced statistical and deep learning methods for
energy production and load forecasting;
 Monte Carlo Methods for Grid disturbance simulations;
 ANN and Super Vector Machine (SVM) classification
algorithm for Measurement errors detection; etc.
Table 1 gives an overview of statistical and Machine
Learning methods applied in recent research and
innovation activities of the Institute Mihajlo Pupin.
D. Artificial Intelligence in Building Operation
Applications
Artificial intelligent techniques can play an important
role in the future of the energy demand response not only
for commercial or public buildings but also for small scale
residential dwellings. Error! Reference source not
found.2, for instance, points to the AI-based pipeline [14]
for residential demand response. The pipeline starts with
collecting sensor measurements, forecasting production
and demand, optimizing energy demand to employing
control actions on the real-world appliances. It provides a
holistic optimization which reschedules end users’ energy
demand in accordance with the available renewable
production and demand response events, but taking into
consideration that demand profile could not significantly
deviate from the forecasted profile.
V. CONCLUDING REMARKS
Smart Grids are cyber-physical energy systems, the next
evolution step of the traditional power grid, and are
characterized by a bidirectional flow of information and
energy. This creates challenges and new software solutions
and hardware components are needed to address the
challenges of optimal energy production, distribution and
consumption. One of the requirements related to data
access procedures in future business solutions for
electricity markets is related to interoperability of energy
services. According to recent EU regulations, energy
services in future will be offered via B2B2C marketplaces
that will enable integration of the energy data in a
European Energy Data Space. Such approach will directly
contribute to quality of life of citizens by making them
active stakeholders in the electricity market chain.
Taking this into account, further research is needed to
achieve adoption of the above mentioned novel concepts
and technologies in the existing proprietary solutions that
are deployed in the Serbian energy value chain.

Fig. 2. AI-powered system for residential demand response [14]
Service
Methods
Non-Intrusive Load
Monitoring [11]
Recurrent Neural Networks (RNNs) with the accent on Long Short Term Memory (LSTM) and
Convolutional Neural Networks (CNNs)
Wind turbine
production forecaster
[12]
Deep neural networks have been chosen for modeling purposes. In order to define number and type of
neural network layers, various architectures have been tested including different combinations of Long-
Short Term Memory (LSTM), Convolutional, Dense and Dropout layers.
Consumption
prediction [13]
Random forests, K-Nearest Neighbor, Support Vector Machines, Linear regression and Neural
networks.
Demand response for
the residential sector
[14]
Auto regressive integrated moving average (ARIMA), linear regression, support vector
regression and K-Nearest Neighbor (kNN).
LITERATURE
[1] European Commission, Powering a climate-neutral economy:
An EU Strategy for Energy System Integration, COM(2020)
299 final, 8.7.2020.
[2] EU Data Strategy European Commission, A European Strategy
for Data (19 February 2020, COM(2020) 66 final).
[3] European Green Deal strategy
[4] M. Makoschitz, et al, “Wide Band Gap Technology: Efficiency
Potential and Application Readiness Map” IEA 4E PECTA
report, May 2020.
[5] T. Strasser, F. Andrén, J. Kathan, et al., “A Review of
Architectures and Concepts for Intelligence in Future Electric
Energy Systems,” IEEE Transactions on Industrial Electronics,
vol.62, no.4, pp. 2424-2438, 2015.
[6] Budka, Kenneth C., Deshpande, Jayant G., Thottan, Marina,
"Communication networks for Smart Grids", Springer Verlag,
2014
[7] J. Stöckl et. All, SINERGY Deliverable 2.1. 2021
[8] European Commission, Proposal for a Regulation of the
European Parliament and the Council on European Data
Governance (Data Governance Act, 25 November 2020,
COM/2020/767 final).
[9] IRENA, Transforming the Energy System and Holding the Line
on Rising Global Temperatures: Key Findings and Numbers,
2019.
[10] “Big Data Analytics in Smart Grids: State-of-the-Art,
Challenges, Opportunities, and Future Directions”, IET
Generation, Transmission and Distribution, February 2019.
[11] Pujić, D., Jelić, M., Tomašević,N., Batić, M, “” Case Study
from the Energy Domain, in
[12] Pujić, D., Janev, V., “Machine learning based wind turbine
production forecaster”, LAMBDA Workshop 2021, CEUR.WS
(in print)
[13] Jelić, M., Pujić, D., Batić, M., “Energy Efficiency
Benchmarking for Smart Homes”, LAMBDA Workshop 2021,
CEUR.WS (in print)
[14] Esnaola-Gonzalez, I.; Jelić, M.; Pujić, D.; Diez, F.J.;
Tomašević, N. “” An AI-Powered System for Residential
Demand Response. Electronics 2021, 10, 693.
https://doi.org/10.3390/electronics10060693
[15] S. Zhao, F. Blaabjerg, und H. Wang, „An overview of artificial
intelligence applications for power electronics“, Juni 2020, doi:
10.36227/techrxiv.12431081.v1.
[16] Shuo Chen, Guomin Zhang, Xiaobo Xia, Yixing Chen, Sujeeva
Setunge, Long Shi. The impacts of occupant behavior on
building energy consumption: A review., Sustainable Energy
Technologies and Assessments, Volume 45, 2021, 101212,
ISSN 2213-1388, https://doi.org/10.1016/j.seta.2021.101212
[17] William O'Brien, Andreas Wagner, Marcel Schweiker, Ardeshir
Mahdavi, Julia Day, Mikkel Baun Kjærgaard, Salvatore
Carlucci, Bing Dong, Farhang Tahmasebi, Da Yan, Tianzhen
Hong, H. Burak Gunay, Zoltan Nagy, Clayton Miller, Christiane
Berger, Introducing IEA EBC annex 79: Key challenges and
opportunities in the field of occupant-centric building design
and operation, Building and Environment, Volume 178, 2020,
106738, ISSN 0360-1323,
https://doi.org/10.1016/j.buildenv.2020.106738
[18] Tareq Abuimara, Brodie W. Hobson, Burak Gunay, William
O'Brien, Michael Kane, Current state and future challenges in
building management: Practitioner interviews and a literature
review, Journal of Building Engineering, Volume 41, 2021,
102803, ISSN 2352-7102,
https://doi.org/10.1016/j.jobe.2021.102803