The Importance of Robust and Reliable Energy Prediction Models:

Next Generation of Smart Meters

Sergio Jurado

, Àngela Nebot

and Francisco Mugica

Soft Computing Research Group at Intelligent Data Science and Artificial Intelligence Research Center,

Universitat Politècnica de Catalunya, Barcelona, Spain

Keywords: Smart Meter, Fuzzy Inductive Reasoning, Energy Modelling, Trading of Renewable Energy.

Abstract: In this position paper a discussion is performed related to the importance of Energy Prediction Models (EPM)

in the context of trading renewable energy in smart grids and the need to develop a new generation of Smart

Meters (SMs) based on edge computing. If the electricity currently produced and consumed in the low voltage

is expected to grow due to installations of PVs and the electrification of the system, we need to incentivise

local energy trading and provide tools to make it possible. Currently, all energy trading mechanisms in the

literature heavily rely on predictions, therefore, inaccuracy would be translated in low profitability of

prosumer’s investments or even worse, disappointment with the local energy markets. To guarantee

robustness and reliability of predictions, we propose Flexible FIR as EPM to be used during the trading

process and to integrate them in a Next Generation of SMs (NGSMs). In this document some reflections about

new functionalities of NGSMs and first steps of a prototype are also addressed.

1 INTRODUCTION

One of the main drivers to reduce the impact of the

climate crisis is changing the status quo of how we

produce, distribute and consume energy. In the last

decades, this is being redefined due to: the inclusion of

renewable energies and distributed generation; new

technologies such as batteries and high-efficient solar

panels; and the way the energy is consumed through

electric vehicles, new energy habits and so on.

All these drivers required a modernization of the

electricity grid and to unlock new mechanisms that

allows more interaction between players and the

electricity grid. This is what is known as Smart Grid

(SG). A SG is an intelligent electrical network used

for improving efficiency, sustainability, flexibility,

reliability and security of the electrical system by

enabling the grid to be observable, controllable,

automated and fully integrated (Smart Grids

European Technology Platform, 2010; US

Department of Energy, 2009). It allows a seamless

and easy connection of distributed energy resources

such as home batteries, prosumers, etc. to the grid.

https://orcid.org/0000-0003-0086-6341

https://orcid.org/0000-0002-4621-8262

https://orcid.org/0000-0003-2843-0427

The domestic penetration of small-scale

renewable resources enables consumers to become

producers of green energy and empowers local

neighbourhoods and communities to collectively

reduce their carbon footprint by trading locally

produced renewable electricity. Thus, enable Local

Energy Trading (LET) is a key milestone to achieve

carbon emissions targets and for a sustainable

scalability of the SG.

Nowadays many energy retailers apply feed-in

tariffs to motivate prosumers to inject their produced

energy. With the rising decentralization of renewable

energy production (Lesser, 2008), it is a challenge to

offer subsidies that ensure a profitable and balanced

grid for all parties involved. There rises the need to

design LET mechanism that aligns the objectives of

individual prosumers, who are aiming for high profits

from their investments, with the objectives of

governments seeking long term positive

environmental change. In addition, with the high

penetration of wind, solar power and customers’

active participation have lead LET to operate in more

uncertain, complex environments. Currently, Energy

248

Jurado, S., Nebot, À. and Mugica, F.

The Importance of Robust and Reliable Energy Prediction Models: Next Generation of Smart Meters.

DOI: 10.5220/0009885802480254

In Proceedings of the 10th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2020), pages 248-254

ISBN: 978-989-758-444-2

Prediction Models (EPM) are used to get the user’s

electricity characteristic curve and required for proper

scheduling activities, power systems planning and

operations, revenue projection, rate design, energy

trading, and so forth. EPMs must be robust and

reliable enough to work under uncertain

consumer/prosumer behaviour and with intermittent

data (missing information), because LET

mechanisms heavily rely on predictions.

This evolution towards a system able to manage

prosumers, batteries, LET and EPMs, in an efficient

and decentralized way, has called for the deployment

of more advanced metering systems. Current Smart

Meters (SMs) aims at monitoring several key

parameters as power quality, remote service switch,

outages, which are helping Distribute System

Operators (DSO) in their load forecast process hence

in a more effective operation of their grid. The use of

SMs helps reduce metering errors and identify fraud,

and reduces the gap between peak demand and the

available power at any given time as well (Council of

European Energy Regulators, 2017).

Nevertheless, most of first generation SMs are

starting to become outdated. Several important and

strictly necessary services, not included in the current

generation of SMs, have to be part of them in order to

be considered smart devices. Having these features is

essential for the evolution towards a system able to

manage prosumers, batteries, LET and EPMs in an

efficient and decentralized way. From our

perspective, this situation has called for the

deployment of a Next Generation of Smart Meters

(NGSMs). We believe that these devices are meant to

orchestrate a set of new functionalities that will bring

Smart Grid goals to a next level.

In this position paper we address the need to

develop a new generation of SMs based on edge

computing that allows not only the prediction of

consumptions and net energy that small producers can

provide to the local grid, but also the maximization of

profits if they participate in the electricity energy

market trading.

The paper is organized as follows: in section 2 an

overview of LET mechanisms in the literature is

presented with focus in the incentive mechanism used

and its dependency in the forecasting. In section 3 we

review the concept of energy models and perform a

literature review of short-term load and production

forecasting techniques that could be embedded in the

aforementioned EPM. In section 4, we present our

opinion about some of the new functionalities that a

NGSMs should have. In section 5 a final discussion

and conclusion is performed on this topic as well as

next steps in our research.

2 LOCAL ENERGY TRADING

MECHANISMS

Trading of locally produced renewable energy is

mainly addressed in literature from a market

perspective and under multiagent based techniques,

where prosumers and consumers (or collectively:

agents) participate in a double auction and trade

energy on a day-ahead basis (Kok, 2005; Kok, 2012;

Olson, 1999; Vytelingum, 2008; Vytelingum, 2010;

Mockus, 2012; Sesetti, 2018; Luo, 2019). Buy and

sell orders for energy are submitted to a public

orderbook and orders are matched either in a

continuous fashion (Vytelingum 2008 and 2010) or at

discrete market closing times using the equilibrium

price (Kok, 2012; Mockus, 2012). The advantages of

this market-based control concept are that it achieves

close to optimal allocation, neatly balances supply

and demand and aligns the preferences of self-

interested agents.

Bidding for energy ahead of time relies heavily on

predictions of future supply or demand, for instance,

in a recent study (Luo, 2019), the LCA (Local

Coordination Agent) performs very short-term

forecasting to predict the power generation and

consumption over future n time interval. Although the

forecasting method itself is not the focus of the study,

it relies on these predictions and the inaccuracy of

which translates to higher costs for both buyers and

sellers. In addition, agents need to rely on advanced

trading strategies in order to maximise profit (or

minimise costs). For example, prosumers with an

inefficient energy forecasting strategy may

unintentionally set a too high sell price, resulting in

an unmatched order for their energy. Since there is no

buyer at the time when they produce and inject the

energy into the grid, prosumers make zero profit,

unless they invest in batteries that can store the

untraded energy. Those agents can then inject the

energy at the time they find a buyer. Lastly, separate

energy balancing mechanisms need to be employed

(Kok, 2012) to cope with real-time demand response.

Another different approach but still using multi-

agents is to consider incentive mechanisms instead of

support policies such as net metering and feed-in

tariff. In (Mihaylov et al., 2014a) it is proposed the

NRG-X-Change, a novel mechanism for trading of

locally produced renewable energy that does not rely

on an energy market or matching of orders ahead of

time. This mechanism uses a new decentralized

digital currency for energy exchange, called

NRGcoin (Mihaylov et al., 2014b). All payments by

consumers and to producers are carried out in

NRGcoins, instead of fiat money. The currency can

The Importance of Robust and Reliable Energy Prediction Models: Next Generation of Smart Meters

249

then be exchanged on an independent open market for

its monetary equivalent, e.g. Euro, Dollar, Pound, etc.

This mechanism is based on a blockchain technology.

Even in market-based mechanisms an EPM is

needed: independently from injection and withdrawal

of energy, NRGcoins are traded on an open currency

exchange market for their monetary equivalent.

Agents use an EPM based on Random Forest (RF)

technique to determine the quantity to trade and the

adaptive attitude bidding strategy to determine the

bid/ask price (Mihaylov et al., 2014b).

3 ENERGY MODELLING

The aforementioned energy trading mechanisms have

in common that they mostly rely in energy models.

An energy model is a computer based model of an

energy system or component, for instance, the

production of a power station or a prosumer, the

consumption load profiles of an entire building or an

appliance, or the behaviour of an entire electricity

distribution system.

Energy models are mainly used for simulations,

which allows us to save resources and time. It allows

us to take into account most of the variables that play

an important role, such as the consumption, weather,

the people, the utility rates and so on. Energy

modelling allows us to represent, analyse, make

predictions, and provide insight into real systems. In

the case of a dwelling for instance, it helps us to

choose between different usage, designs and

materials. By adjusting variables, we can check their

impact on the energy requirement. There are many

different energy models and applications;

backcasting models, scenario analysis models,

integrated models, demand/supply-side models, etc.

(Farzaneh, 2019). For energy trading purposes

demand and supply-side models are needed.

 Demand-Side Models: These consist of a broad

range of methodologies which focus on

determining the final energy consumption in the

entire economy or a particular sector, such as the

buildings, industrial energy use, and the

transportation system. The overall

methodological focus of this cluster of energy

system models is to consider the demand side

endogenously, and the supply-side issues are not

considered at all. These models mostly rely on

bottom-up simulation techniques to estimate

energy demand.

 Supply-Side Models: Mostly focused on energy

supply technologies, with a particular focus on

renewable energy systems, fossil-based power

plants, oil and gas industries, etc. They are

characterized by a limited spatial scale and

generally consider a single piece of technology

using a simulation technique or experimental

work to perform the analysis, including the design

and performance of the system. The models may,

therefore, be characterized as calculating supply-

side parameters related to technology design or, in

some cases, the operation of such technologies.

Electricity load and production forecasting is

typically divided in short, medium and long term. The

long-term plan evaluates how well the short-term

planning commitments fit into long-term needs. No

commitment needs to be made to the elements in a

long-term plan, and capacity and location are more

important than timing in long-term forecast. In other

words, it is more important to know what will

eventually be needed than to know exactly when it

will be needed. Each category is equally important in

the energy sector for the correct operation of the

power system. For the purpose of this paper the mid

and long term are not considered because the trading

between prosumers and consumers is generally in

hourly basis or less.

Sort-term Load and Production Forecasting

(SLPF) is highly connected with meteorological

factors such as temperature, humidity, wind speed

and specially typology of day. The change in

holidays, weekdays, weekends, the day before and

after holidays also has impacts on the load forecast

(Jurado et al., 2015). The analytical methods work

well under normal daily circumstances, but they can’t

give contenting results while dealing with

meteorological, sociological or economical changes,

hence they are not updated depending on time. On the

contrary, AI techniques have indicated the capability

of learning complex nonlinear relationships, which

are difficult to model, and accordingly making them

popular (Jurado et al., 2015).

There are some State-of-the-Art (SoA) review of

AI load forecasting approaches classification and

comparison. Hong (2016) offers an impressive review

of most important SLPF literature over the last forty

years divided by conceptual and empirical studies.

Moreover, the article includes a review of most

notable techniques, i.e. ANN, Fuzzy Logic, SVM,

Gradient Boosting, evaluation methods, and common

misunderstandings. Another SoA empirical review of

three AI techniques; ANN, SVM and Adaptive

Neuro-Fuzzy Inference System (ANFIS) is

performed by (Zor, 2017). Studies investigated in the

context of this paper show that three different AI

techniques have the potential for excellent

forecasting. Another conceptual SoA review is

SIMULTECH 2020 - 10th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

250

performed by (Singh, 2012), where they classify

demand forecasting techniques in i) traditional

mathematical techniques i.e., Regression, Multiple

Regression, Exponential Smoothing, etc.; ii)

Modified Traditional techniques i.e., Adaptive

Demand Forecasting, AR, ARMA and ARIMA

model, SVM; and iii) SC techniques such as Genetic

Algorithms (GA), ANN, Fuzzy-Logic and

Knowledge-Based Expert Systems.

An implementation of these technologies for a

massive deployment and/or for its usability in other

processes such as in the LET, features such as

robustness and reliability are an essential. The

information that arrives from the different sensors in

the home area network and/or the SM, have problems

that may hinder or even prevent the forecasting and

hence an optimal electricity trading.

Jurado et al. (2017) have done important steps in

this direction. Flexible FIR is an improved version of

FIR, a hybrid methodology based on fuzzy logic and

inductive reasoning, which has been demonstrated to

predict under scenarios of high number of missing

values and therefore, uncertainties. Moreover,

Flexible FIR uses a kNN optimal selection algorithm

(Jurado et al., 2019) during the FIR prediction phase

that allows the model to select the most suitable

number of nearest neighbour, improving the accuracy

and almost without impact in the model parameter

selection.

These are remarkable feature because it could

help for instance, in energy trading where missing

data is present and using hardware with limited

specifications.

Additional information about FIR, Flexible FIR

and its applications in the energy domain and

experiments performed can be found in (Jurado et al.

2013, 2015, 2017, 2019)

4 EDGE COMPUTING AND A

NEXT GENERATION OF

SMART METERS

Every home in Europe should be offered a SM from

their energy supplier in the next few years. The

government of Britain, for example, has pledged to

offer all households the option of a SM by 2024

(Uswitch, 2020).

The main advantages of SM with respect

traditional meters from final customer perspective

are:

 Increase bills accuracy, since no more

estimations and no more meter readings are

necessary;

 Reduce the problems when customers switch

suppliers. The SM newer models are

compatible with the network that the meters

talk to all suppliers through.

 Update readings frequently enough to use

energy savings schemes;

 The in-home display enables the user to see

how much energy is using at different times of

the day. Some of them include a user-friendly

App that shows current energy consumption,

total balance and compares the usage

performed in different months and years;

 Reduce energy bills, up to some extend. With

the information that the in-home display offers

(previous point), the consumer can try to cut or

modify the energy usage and, therefore, be

more energy efficient;

 Increase customer’s energy usage

understanding. The SM allow the user to see

the direct impact of the family habits on the

bill.

And from a DSO/reatiler point of view (Prettico,

2019):

 Allow remote reading by the operator;

 Provide 2-way communication for

maintenance and control;

 Allow frequent enough readings to be used for

network planning;

 Support advanced tariff schemes;

 Allow remote ON/OFF control of power

supply and/or flow or power limitation;

 Provide secure data communication;

 Allow fraud detection and prevention;

 Provide import/export and reactive metering.

More details about energy meters evolution in

smart grid can be found in (Avancini, 2019).

SMs are part of the effort to create a smart grid,

which is part of providing low-carbon, efficient and

reliable energy to households. Households will find

themselves in a position to feed supply back into the

grid, as well as draw from it (Logic4Training, 2020).

Therefore, trading of locally produced renewable

energy becomes fundamental.

However, from our point of view, and taking into

account the current services offered, the so called

“smart” meters cannot be considered smart at all.

Several important and strictly necessary services and

features, not included in the current generation of

SMs, have to be part of them in order to be considered

smart devices. We believe in edge computing as a

The Importance of Robust and Reliable Energy Prediction Models: Next Generation of Smart Meters

251

core technology of the NGSMs. Edge computing was

detected as a Top 10 Strategic Technology Trends for

2018 (Cearley, 2018) and since then, the number of

applications and sectors where it has been applied has

increased exponentially. Edge computing delivers the

decentralized complement to today’s hyperscale

cloud and legacy data centres. Edge computing

addresses the limitations of centralized computing

(such as latency, bandwidth, data privacy and

autonomy) by moving processing closer to the source

of data generation, “things” and users. To maximize

the applications potential and user experience, DSOs

need to plan distributed computing solutions along a

continuum from the core to the edge.

From our point of view, the features that the

NGSMs should provide are:

 Individual electricity load and production

forecasting: With an EPM in each Smart Meter

utilities would have information in each local

node of the network, which means visibility

and predictability in the Low Voltage.

Moreover, the prediction at local level would

be used to enable other features such as

optimization of the energy trade and demand

response programs.

 Trading mechanisms to allow peer-to-peer

energy trading: Agent-based technology with

multiple incentive mechanisms for rewarding

renewable energy production and

consumption.

 Demand-side management of home devices for

demand response programs: Exploit and

manage domestic consumer policies using the

Internet of Things (IoT). Turn on/off high

consumption equipment in the best price bands.

Detect consumption that have been made

outside the planned hours.

 Load disaggregation: This function would,

among other things, allows to analyse and

detect household appliances in poor conditions,

appliances that consume a lot of energy or

appliances and devices that have been left on.

Along with the ability to interact with

appliances (IoT), the smart meter could deploy

a consumption optimization policy.

 Security and privacy by integrating blockchain

technologies: When some sort of trading occurs

between multiple parties, trust is a major

concern. Traditionally, a third part trusted by

all keeps the record of transactions. Using

blockchain technology for energy trading

eliminates the role of trusted third party.

 5G connectivity: Interconnecting IoT with 5G

networks efficiently helps to manage energy

balance. This helps in the reduction of energy

cost. Efficient data analysis can be done

through 5G networks, which could empower

cities to execute their own energy plans that

would be more cost effective in accordance

with demographic conditions. Moreover, 5G

networks largely help the different distribution

operators to reach their observability down the

substation level. This assures substantial and

balanced operations in the grid.

In Figure 1 we propose a high level architecture

of a NGSM. The architecture is based on a Bourns

SM.

Figure 1: High level architecture of a NGSM.

The three main internal areas of a SM design

include the power system: it has a switched mode

power supply and battery backup to ensure that the

metering electronics remain powered;

MicroController Unit (MCU): typically includes an

Analog-to-Digital Converter (ADC), Digital-to-

Analog Converter (DAC) and in blue tones we have

added the proposed components in the architecture

that would allow the features previously mentioned;

and finally, communications interface (HAN, WAN

and Optional Interfaces): a wired or wireless

communication interface allows the meter to interact

with the rest of the grid, and in some cases the end

user’s network.

The new architecture is inspired in the edge

architecture of IBM (Iyengar, 2019). The NGSM is

equipped to run analytics, apply AI rules, and even

store some data locally to support operations at the

Edge. The NGSMs could handle analysis and real-

time inferencing without involvement of the Edge

server or enterprise layer. This is possible because

devices can use any Software-as-a-Service (SaaS).

Driven by economic considerations and form factors,

an Edge device typically has limited compute

SIMULTECH 2020 - 10th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

252

resources. It is common to find Edge devices with

ARM or x86 class CPUs with one or two cores, 128

MB of memory, and perhaps 1GB of local persistent

storage

Having these features is essential to be able to

trade renewable energy on smart grids. Some steps in

this direction has been already done by some of the

authors of this paper. Under the demonstrations in

simulation environments of the NRG-x-Change and

the NRGCoin concepts (Mihaylov et al., 2015) a

prototype of SGSM was created, using Raspberry Pi

to unlock edge computing capabilities. Prosumers

were represented as software agents running on

individual Raspberry Pi boards and 56 consumers’

agents where running in individual threads on two

Raspberry Pi boards. NGSMs were connected to the

Internet, which allows agents to submit orders for

buying and selling NRGcoins, allowing agents to

trade energy through their SFSMs. Orders were

matched in real-time using continuous double

auction, as employed by the New York Stock

Exchange. All software agents were developed in

Java and implemented inside the Raspberry Pi, while

the exchange market was developed in C# using

Azure Service Bus for synchronizing actions. All

components communicated using the RESTful

Microservice architecture.

Agents used an EPM based on RF technique to

determine the quantity to trade and the adaptive

attitude bidding strategy to determine the bid/ask

price. We are currently working to implement agents

based on Flexible FIR, which has been proven in

other studies to predict in a more robust and reliable

way electricity load of multiple characteristic curves.

With this new implementation we believe we will be

able to improve negotiations between agents which

will be translated in a major profitability of customer

investments in renewable energy.

This implementation is being done using

Raspberry Pi (Raspberry Pi Foundation), which

simulates a NGSMs. We are aware that this hardware

is not compiling with industry standards, however,

the communication protocols, operating system

(open-source software implementations for the agent,

trading strategies and EPM) CPU, memory and

storage are hardware and software requirements

easily integrable.

We are now working with EPMs based on

different techniques; ANN, RF, ARIMA and Flexible

FIR, and how the performance in predictions directly

affects to the objectives of individual prosumers.

5 DISCUSSION AND

CONCLUSION

We believe new generation of SMs must provide

citizens new ways to interact with the energy markets

and services that helps the society to achieve the

challenging energy goals we have in the next 20

years. SMs are in a unique position to technologically

enable new features such as energy trading strategies

for a local peer-to-peer in neighbourhoods and

communities. However, if we want to increase local

production and allow for LET, we need new hardware

and software implementations. We believe that this

has to be addressed from a decentralized point of

view, for example with edge computing in a SGMSs.

A key component inside SGMS will be the EPM,

because it has to provide not only accurate predictions

to the DSO but also robust and reliable forecasting for

the individual (agent) participating in a local energy

markets to achieve an optimal solution. to the agent

collaborating in a local energy market.

A first proof of concept of the NGSMs has been

implemented, however, the EPM used is far from

being an optimal solution. We propose to use Flexible

FIR for the EPM. RF has been already applied in this

hardware. Our proposal is to bring the negotiation to

a next level by implementing Flexible FIR. Its

performance has been demonstrated for different

consumption profiles and can cope with missing

information in the input values, as well as during the

prediction phase. Moreover, it works well in an

“isolated” approach like in edge computing scenarios.

It does not rely on deep learning or high

computational cost plus it has been demonstrated to

seemly choose autonomously some FIR input

parameters.

It is important to highlight that there will be

processes managed through cloud computing apps

and others must be considered. However, we see

cloud computing and edge computing

complementary, rather than competitive or mutually

exclusive. We believe that organizations that use

them together will benefit from the synergies of

solutions that maximize the profitability of both

centralized and decentralized models.

REFERENCES

Avancini, D. B., Rodrigues, J. J., Martins, S. G., Rabêlo, R.

A., Al-Muhtadi, J., and Solic, P., 2019. Energy meters

evolution in smart grids: A review. Journal of cleaner

production, 217, pp. 702-715.

The Importance of Robust and Reliable Energy Prediction Models: Next Generation of Smart Meters

253

Council of European Energy Regulators, 2017. CEER

Report on Power Losses.

CeArley, D., Burke, B., Searle, S., & Walker, M. J. (2016).

Top 10 strategic technology trends for 2018. The Top,

10.

Farzaneh, H., 2019. Energy Systems Modeling: Principles

and Applications. Springer.

Hong, T., and Fan, S., 2016. Probabilistic electric load

forecasting: A tutorial review. In International Journal

of Forecasting, 32(3), pp. 914-938. [Online] [accessed

18.05.20].

Iyengar, A. 2019. Architecting at the Edge. IBM Cloud.

Jurado, S., Nebot, À., & Mugica, F., 2019. K Nearest

Neighbour Optimal Selection in Fuzzy Inductive

Reasoning for Smart Grid Applications. In 2019 IEEE

International Conference on Fuzzy Systems (FUZZ-

IEEE). pp. 1-6.

Jurado, S., Nebot, À., Mugica, F., and Mihaylov, M., 2017.

Fuzzy inductive reasoning forecasting strategies able to

cope with missing data: A smart grid application. In

Applied soft computing, 51, pp. 225-238.

Jurado, S., Nebot, A., Mugica, F. 2015. Hybrid

methodologies for electricity load forecasting: Entropy-

Based Feature Selection with Machine Learning and

Soft Computing Techniques, in Energy, Elsevier, vol.

86, pp. 276-291.

Jurado, S., Peralta, J., Nebot, A., Mugica, F., & Cortez, P.

(2013, July). Short-term electric load forecasting using

computational intelligence methods. In 2013 IEEE

International Conference on Fuzzy Systems (FUZZ-

IEEE) (pp. 1-8). IEEE.

Kok, J., Warmer, C., and Kamphuis, I., 2005. Power-

Matcher: multiagent control in the electricity

infrastructure. In Proceedings of the fourth

international joint conference on Autonomous agents

and multiagent systems, pp 75–82.

Kok, K., Roossien, B., Macdougall, P., and Pruissen, O. V.,

2012. Dynamic Pricing by Scalable Energy

Management Systems - Field Experiences and

Simulation Results using PowerMatcher. In IEEE

Power and Energy Society General Meeting, pp 1–8.

Lesser, J. A., Su, X., 2008. Design of an economically

efficient feed-in tariff structure for renewable energy

development. Energy Policy, 36(3), pp 981 - 990.

Logic4Training. Second Generation Smart Meters are

Coming. [Online] [accessed 04.04.20].

Luo, F., Dong, Z. Y., Liang, G., Murata, J., and Xu, Z.,

2018. A distributed electricity trading system in active

distribution networks based on multi-agent coalition

and blockchain. In IEEE Transactions on Power

Systems, 34(5), pp. 4097-4108.

Mihaylov, M., Jurado, S., Avellana, N., Razo-Zapata, I.,

Van Moffaert, K., Arco, L., and Nowé, A., 2015.

Scanergy: a scalable and modular system for energy

trading between prosumers. In Proceedings of the 2015

International Conference on Autonomous Agents and

Multiagent Systems. pp. 1917-1918.

Mihaylov, M., Jurado, S., Van Moffaert, K., Avellana, N.

and Nowé, A., 2014a. NRG-X-Change: A Novel

Mechanism for Trading of Renewable Energy in Smart

Grids, in Proc. of the 3rd International Conference on

Smart Grids and Green IT Systems (SmartGreens). pp.

101-106.

Mihaylov, M., Jurado, S., Avellana, N., Van Moffaert, K.,

de Abril, I. M., and Nowé, A., 2014b. NRGcoin: Virtual

currency for trading of renewable energy in smart grids.

In 11th International conference on the European

energy market (EEM14). pp. 1-6.

Mockus, J., 2012. On Simulation of the Nash Equilibrium

in the Stock Exchange Contest. Informatica, 23(1), pp

77–104.

Olson, M. A., Rassenti, S. S. J., Smith, V. L., Rigdon, M.

L., and Ziegler, M. J., 1999. Market design and

motivated human trading behavior in electricity

markets. In Proceedings of the 32nd Annual Hawaii

International Conference on System Sciences, pp 1–27.

Prettico, G., Flammini, M. G., Andreadou, N., Vitiello, S.,

Fulli, G., Masera, M., 2019. Distribution System

Operators observatory 2018 - Overview of the

electricity distribution system in Europe, EUR 29615

EN, Publications Office of the European Union,

Luxembourg.

Raspberry Pi Foundation, Raspberry Pi 4 Tech Specs,

[Online] [accessed 04.04.20].

Sesetti, A., Nunna, H. K., Doolla, S., and Rathore, A. K.,

2018. Multi-agent based energy trading platform for

energy storage systems in distribution systems with

inter-connected microgrids. In 2018 IEEE Industry

Applications Society Annual Meeting (IAS), pp. 1-8.

Singh, A. K., Khatoon, S., Muazzam, M., and Chaturvedi,

D. K., 2012. Load forecasting techniques and

methodologies: A review. In 2nd International

Conference on Power, Control and Embedded Systems

pp. 1-10.

Smart Grids European Technology Platform, 2010.

Strategic deployment document for Europe's electricity

networks of the future [Online] [accessed 04.04.20].

US Department of Energy, 2009. Smart grid system report,

July [Online] [accessed 05.04.2020].

Uswitch. Smart meters explained. [Online] [accessed

04.04.20].

Vytelingum, P., Ramchurn, S., Voice, T. D., Rogers, A.,

and Jennings, N., 2010. Trading agents for the smart

electricity grid. In 9th Int. Conf. on Autonomous Agents

and Multiagent Systems (AAMAS 2010), pp 897– 904.

Vytelingum, P., Cliff, D., and Jennings, N., 2008. Strategic

bidding in continuous double auctions. Artificial

Intelligence, 172(14), pp 1700–1729.

Zor, K., Timur, O., and Teke, A., 2017. A state-of-the-art

review of artificial intelligence techniques for short-

term electric load forecasting. In 6th International

Youth Conference on Energy (IYCE). pp. 1-7.

SIMULTECH 2020 - 10th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

254