Manglar Living Lab: Energy Management Through a Smart

Microgrid with Artificial Intelligence

Gustavo García

, Alejandro Guerrero

and Javier E. Sierra

Universidad de Sucre, Sincelejo, Colombia

Keywords: Energy Management, Microgrid, Artificial Intelligence, Sustainable Energy Solutions.

Abstract: Latin America, and Colombia in particular, are making strides in their energy transition by implementing

innovative projects that prioritize sustainability and efficiency. This article presents a conceptual framework

of the Manglar Living Lab pilot plant, detailing the microgrid architecture with the goal of overcoming energy

challenges and focusing on efficient energy management. Central to this initiative is the development of a

smart metering device, driven by artificial intelligence (AI), with a technological platform and real-time

monitoring capabilities. This Living Lab not only bolsters Colombia’s energy transition strategy but also

illustrates the potential of localized AI-powered solutions to enhance energy efficiency and grid reliability in

the region.

1 INTRODUCTION

Living Labs have emerged as innovative platforms

that facilitate co-creation and experimentation in real-

world environments, promoting energy efficiency

and sustainability. These are collaborative spaces

where end-users, researchers, and companies co-

design and test innovative solutions in practical

settings. In the field of energy efficiency, Living Labs

enables the implementation and evaluation of

sustainable technologies and practices in

communities and buildings by Almirall et al. (2012).

Various studies have explored how AI can enhance

the efficiency and stability of microgrids.

The integration of smart technologies, such as

energy management systems and IoT devices, is

essential in energy efficiency-oriented Living Labs.

Ballon et al. (2018) and Mohamed et al. (2016)

analyze how these technologies facilitate real-time

monitoring and optimization of energy consumption

in urban environments, thereby enhancing

sustainability and reducing costs.

Active participation of end-users is crucial in

Living Labs. Leminen et al. (2012) explore how co-

creation and community involvement in the design

and testing of energy solutions lead to greater

https://orcid.org/0009-0003-0248-7844

https://orcid.org/0000-0001-5179-6473

https://orcid.org/0000-0002-9489-0520

acceptance and effectiveness of implemented

initiatives.

Several university institutions have implemented

Living Labs to promote energy efficiency:

● The Smart City Málaga project is designed to

transform university campuses into smart cities,

facilitating the efficient management of

resources while fostering innovative research

and educational activities. These factors are

crucial to the development of future smart cities.

The project's emphasis on energy efficiency

measures and active demand management

strategies led to a significant reduction in energy

consumption (Forte at all. 2019).

● The University of Genoa has established the

"Living Lab Smart City" to test cutting-edge,

sustainable technologies for energy production,

distribution, and management. The goal is to

transform the Savona Campus, which resembles

a small urban district with approximately 2,500

residents, into a model of an innovative and

sustainable city by implementing demonstrative

infrastructure that can be replicated at the city

district level (Laiolo et al.2021).

García, G., Guerrero, A. and Sierra, J. E.

Manglar Living Lab: Energy Management Through a Smart Microgrid with Artiﬁcial Intelligence.

DOI: 10.5220/0013275500003953

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 14th International Conference on Smart Cities and Green ICT Systems (SMARTGREENS 2025), pages 99-105

ISBN: 978-989-758-751-1; ISSN: 2184-4968

● Akpolat et al. (2024) proposes the planning,

modeling, implementation, and operation of a

laboratory-scale distributed energy resource

(DER) and a living laboratory structure with a

hybrid energy system. This system utilizes

photovoltaics, a small-scale wind turbine, a

proton exchange membrane fuel cell, and a lead

acid battery energy storage system. The

objective of this system is to enhance the

standards of education and research within the

electrical and electronic engineering fields.

Specifically, this system structure has facilitated

opportunities for researchers and students to

engage in diverse areas such as renewable

energy, control systems, power electronics,

energy management systems, and software

development.

● The Sorocaba Institute of Science and

Technology (ICTS) of Unesp intends to initiate

its transition into a smart campus. The first step

in this endeavor is to adopt measures related to

the measurement of consumption and control of

electric energy and water, thereby establishing a

solid foundation for the implementation of TIC

(Zarpellon, B. O. 2024).

Despite the benefits, the implementation of Living

Labs faces challenges such as management of the data

generated and long-term sustainability. Schuurman et

al. (2016) address these challenges and propose

strategies to overcome them, emphasizing the

importance of proper planning and continuous

adaptation to community needs.

1.1 Integration of Artificial Intelligence

(AI)

The implementation of smart grids, which combine

traditional electrical grids with modern information

and communications technologies, as well as

distributed generation systems and microgrids, is no

longer a futuristic concept but a reality in many

developed nations. The emergence of AI in managing

the energy use of smart microgrids has presented an

auspicious solution for optimizing energy distribution

and consumption. This is particularly relevant given

the increasing integration of renewable energy

sources.

Bernstein et al. (2015) and Mastelic et al. (2017)

proposed a composable method for real-time control

of active distribution networks by utilizing explicit

power setpoints. This approach effectively

coordinates distributed resources, enhancing both

grid stability and efficiency. Similarly, Simões and

Farret (2014) emphasized the application of artificial

intelligence (AI) techniques, such as neural networks

and fuzzy logic, in renewable energy systems and

microgrids. Their research demonstrates how these

methods can optimize microgrid operation and

control, adapting dynamically to fluctuations in

energy generation and demand. In Le, T. T. H., &

Kim, H. (2018) the research consisted of detecting

home appliance events using low frequency active

power signals. Decision Tree and Long Short-Time

Memory (LSTM) were used to detect ON/OFF events

with high accuracy reaching accuracies of 92.64%

and 96.85% respectively.

Lazzarett et al. (2020) propose load monitoring

modules that detect power changes, identify loads,

and report data to an operations center. The detection

method is based on half-cycle apparent power and

signature analysis using power envelopes.

Da Silva Nolasco. (2021) proposes a new

architecture based on "Convolutional Neural

Network" CNN that integrates detection, feature

extraction and classification of high frequency NILM

signals. Their approach uses a grid to divide the input

signal into segments and a series of convolutional

layers to extract features from the signal.

Demonstrating greater feasibility to be used in the

real world.

Mohseni et al. (2022) introduced a capacity

planning model for off-grid microgrids, utilizing

metaheuristic-based optimization algorithms. This

model proved to be effective in planning and

managing microgrid operations in remote areas,

significantly improving energy efficiency and

reducing costs.

The implementation of the Living Lab with the

integration of non-intrusive load monitoring (NILM)

can become an emerging approach with cost-effective

energy management solutions. This is achieved by

utilizing the aggregate load obtained from a single

smart meter within the power grid. Furthermore, by

integrating machine learning (ML), NILM can

efficiently utilize electrical energy and lessen the load

for the energy monitoring process (Silva, M. D., &

Liu, Q. (2024)).

These studies highlight the potential of AI to

revolutionize energy management in smart microgrids,

providing more efficient and adaptive solutions to the

current challenges in the energy sector.

1.2 Our Contribution

Latin America, and particularly Colombia, is actively

pursuing projects to address the global energy

transition challenge by integrating innovative

SMARTGREENS 2025 - 14th International Conference on Smart Cities and Green ICT Systems

100

technologies and sustainable practices. This article

introduces a groundbreaking Living Lab developed in

Colombia as part of these efforts, focusing on energy

efficiency and intelligent energy management. The

Living Lab, named MANGLAR, serves as a

collaborative space for the co-creation and real-world

testing of advanced energy solutions.

The centrepiece of this initiative is a novel

technological platform and smart measuring device,

designed to optimize the operation of a microgrid.

Leveraging artificial intelligence (AI), the platform

enables real-time monitoring, demand-response

optimization, and predictive analytics, ensuring

efficient energy use and improved grid reliability. This

integration of AI into the microgrid management

framework represents a significant advancement in

addressing the region’s energy challenges, particularly

in decentralized and renewable-based power systems.

The article presents the conceptual framework of

the Living Lab, details the architecture of the smart

microgrid, and highlights the results from pilot

implementations. By contextualizing this innovation

within Colombia’s broader energy transition strategy,

the work underscores the importance of localized

solutions and technological innovation in achieving

sustainable energy goals in Latin America.

Next, section 2 then provides a detailed

description of the generation system available in the

Living Lab. It also focuses on the equipment and tools

that enable real-world experimentation. Finally,

Section 3 presents the design of a smart metering

device that employs edge AI to efficiently manage the

microgrid. This work highlights the importance of

localized solutions and technological advancements

as key enablers of Colombia’s energy.

2 SUSTAINABLE ENERGIES:

ADVANCING SOLAR

GENERATION

In Colombia, the electricity service is stratified

according to family income and work activities. The

regulated electricity market comprises sectors 1 to 3,

which include low-income households; sector 4

belongs to middle-income households. Sectors 5 and

6 are made up of higher-income households and the

commercial sector. On the other hand, the industrial

sector belongs to the unregulated market, where they

can choose the marketer and freely agree on the price

of electricity (Salazar G. (2013)). In Guerrero

Hernández et al. (2022) a mixed integer optimization

model is proposed for energy management in the

Manglar Lab under the conditions of the tariff system

in Colombia.

The name Manglar had its origin in a coastal

ecosystem where unique trees and shrubs thrive,

swampy where solar radiation and winds are excellent

for the generation of electricity. These unique trees,

through the process of photosynthesis, capture solar

energy and become a support and food habitat for

many marine and terrestrial species. Manglar Living

Lab was created in the Energy 2030 project, financed

by the World Bank. This electronic engineering

research laboratory at the University of Sucre

becomes a fundamental pillar in the energy transition

of the university campus and a place for the

generation of knowledge in the areas of power

electronics, artificial intelligence and device design.

The Manglar Living Lab has a capacity of 20 kWp

in the solar generation system, subdivided into three

sections, each composed of 8 JKM400 reference

photovoltaic panels. In addition, it has three CPS

SCA6KTL-SM type inverters and a BESS storage

system with an energy capacity of 12.6 kWh, which

provides electrical energy to the laboratory in case of

absence of the conventional electrical supply (see

figure 1).

Figure 1: Microgrid Manglar.

The building has four floors with an elevator and solar

panels installed on the roof. This building houses

physics, chemistry, agribusiness and electronics

laboratories. It has an average daily consumption of

5903kWh. Figure 2 shows the data of the power

generated by the group of 8 solar panels connected to

an inverter in a measurement period of 48 hours.

Figure 2: Data obtained from the solar inverter.

Manglar Living Lab: Energy Management Through a Smart Microgrid with Artiﬁcial Intelligence

101

The laboratory is equipped with a lithium battery

bank that provides up to 2 hours of electrical

autonomy. The importance of a battery bank in a

microgrid is pivotal to achieving efficient energy

management and cost optimization, particularly in

scenarios where the aim is to minimize operational

costs Mazidi el al (2023). This physical space houses

equipment such as 5 personal computers, a modem, a

3D printer, a video wall, an LED lamp, a water

dispenser, and an electric coffee maker (see Table 1).

Table 1: Electronic devices in the laboratory.

Quantity Apparatus Power (W)

1 Water Dispense

1 Electric Coffee Make

900

1 3D Printe

350

1 Route

4 Led Lam

4 Video wall 250

5 PC 340

1 CNC 200

2.1 Smart Microgrid

Figure 3 illustrates an energy management system for

"Manglar Living Lab". The building icon with solar

panel and battery represents a solar energy system

with energy storage. The thick black vertical line

symbolises the main source of energy that powers the

laboratory, and the surplus is distributed throughout

the building. The enclosure designated as "NILM"

serves as a non-intrusive load monitoring system.

This advanced technology facilitates the

identification of energy consumption associated with

individual devices without the necessity of installing

separate meters for each one. It operates by analysing

fluctuations within the total electrical current, thereby

disaggregating the consumption of each appliance.

An Emporia® Gen 3 energy meter is used to

record laboratory consumption. The average daily

consumption is 33.5 kWh and figure 4 shows the

average daily load profile.

3 SMART MEASURING DEVICE

This section describes the prototype for monitoring

and managing electric energy using artificial

intelligence (AI). This system, composed of various

interconnected components, collects real-time data on

the electric flow and then analyses it to make

decisions that improve the efficiency and stability of

the network.

Figure 3: Energy management system for "Manglar Living

Lab".

Figure 4: Average daily laboratory consumption.

Figure 5: Integration of ADE9000 and jetson Orin.

Firstly, current and voltage sensors, located at

strategic points in the electrical network, capture

crucial information about the magnitude of the

electrical flow. This data is transmitted to the

ADE9000, an integrated circuit specializing in the

SMARTGREENS 2025 - 14th International Conference on Smart Cities and Green ICT Systems

102

measurement of energy variables such as RMS

voltage, RMS current, active power, and power

factor, among others. This device acts as a collection

and preprocessing center, and in turn, sends the

processed information to the Jetson Orin, a powerful

NVIDIA computing platform with 8GB of RAM and

an Arm® Cortex-A78AE v8.2 6-core processor, via

the SPI communication protocol (see figure 5).

The Jetson Orin is the brain of the system, where

the AI algorithms responsible for analysing the

received data reside. These algorithms, trained to

recognize patterns and anomalies, can detect fluctua-

tions in demand, identify possible network failures,

and even predict future energy needs in real time. This

analytical capability allows the system to make

intelligent decisions to optimize energy distribution,

balance the network load, and ensure a stable supply.

Finally, the Jetson Orin, through a Wi-Fi

connection, communicates with a cloud-based

management system. This connection allows remote

monitoring of the network, storage of historical data

for long-term analysis, and updating of AI algorithms

with new information (see figure 6). Data is

visualized and stored in InfluxDB.

Figure 6: Average daily laboratory consumption.

Figure 7 depicts the fluctuation of voltage,

current, and power within the Manglar living lab's

electrical system over time. There is a clear

correlation between the three variables: an increase in

current corresponds with an increase in power, and a

slight decrease in voltage. This suggests that the

system is subjected to a variable load, and that voltage

can be affected when energy demand is high. Also, in

8 displays the fluctuation of frequency and power

factor in an electrical system over time. The

frequency, although it remains close to 59.95 Hz,

presents variations that, although they appear small,

can indicate instabilities in the generation or

distribution of electrical energy. A fluctuating power

factor such as the one observed in the second chart,

with values ranging between 0.6 and 0.9, suggests

that electrical energy is not being used efficiently. It

is important to monitor these fluctuations, since they

can affect the performance and useful life of the

equipment, as well as increase energy consumption

and associated costs.

Therefore the Artificial intelligence (AI) presents

a range of possibilities to optimize instabilities in

electrical loads and the identification of laboratory

equipment. Potential strategies include:

• Optimizing energy use: AI can identify

consumption patterns and recommend strategies

to optimize energy use, such as scheduling the

operation of high-consumption equipment

during periods of low demand or implementing

automatic shutdown systems.

• Responding to unforeseen events: AI can detect

anomalies in consumption and react

autonomously to unforeseen events, such as

demand peaks or supply failures, by activating

mitigation measures to minimize the impact on

loads.

• Machine learning platforms: These platforms

allow for the development and training of AI

models for prediction and control purposes.

Figure 7: Average voltage, current and power variables

with the prototype.

Manglar Living Lab: Energy Management Through a Smart Microgrid with Artiﬁcial Intelligence

103

Figure 8: Frequency and power factor measurement with

prototype.

4 CONCLUSIONS

The MANGLAR Living Lab, equipped with a smart

microgrid prototype and AI technology, is being

tested as a tool for managing electrical energy in

laboratory environments. The results obtained so far

are visualized and stored in InfluxDB. Storing this

data will allow us to understand the behavior of the

load profile and perform AI training for energy

management. It also helps us predict future energy

needs, which facilitates decision-making and

resource optimization.

Despite the progress made, the MANGLAR

project is in its initial phase and presents

opportunities for future research. A more

comprehensive evaluation of the long-term impact of

the intelligent energy management system is

required, including cost-benefit analysis and

scalability assessments. Furthermore, the integration

of new renewable energy sources, such as wind or

biomass, and the development of more sophisticated

AI algorithms could further improve the efficiency

and resilience of the microgrid.

ACKNOWLEDGEMENTS

The authors would like to thank the Research

Program "Energy Efficiency 2030: Transition to

Sustainable Construction”, with code 1216-938-

106387, funded by the Ministry of Science,

Technology and Innovation (Minciencias) of the

Government of Colombia through the call "938-2023

Ecosystems in Sustainable, Efficient and Affordable

Energy" with contract No. 395-2023.

REFERENCES

Almirall, E., Lee, M., & Wareham, J. (2012). Mapping

Living Labs in the Landscape of Innovation

Methodologies. Technology Innovation Management

Review, 2(9), 12-18.

Akpolat, A. N., Dursun, E., & Kuzucuoğlu, A. E. (2024).

Modeling and operation of a fuel cell stack for

distributed energy resources: A living lab platform.

International Journal of Hydrogen Energy

Ballon, P., Pierson, J., & Delaere, S. (2018). Open

Innovation Platforms in Smart Cities: An Inter-

Organizational Perspective. Computer, 51(6), 60-67.

Bernstein, A., Reyes-Chamorro, L., Le Boudec, J.-Y., &

Paolone, M. (2015). A composable method for real-

time control of active distribution networks with

explicit power setpoints. Electric Power Systems

Research, 125, 254-264.

Da Silva Nolasco, L., Lazzaretti, A. E., & Mulinari, B. M.

(2021). DeepDFML-NILM: A new CNN-based

architecture for detection, feature extraction and multi-

label classification in NILM signals. IEEE sensors

journal, 22(1), 501-509.

Fortes, S., Santoyo-Ramón, J. A., Palacios, D., Baena, E.,

Mora-García, R., Medina, M., ... & Barco, R. (2019).

The campus is a smart city: University of Málaga

environmental, learning, and research approaches.

Sensors, 19(6), 1349.

Guerrero Hernández, A. S., & Ramos de Arruda, L. V.

(2022). Technical–economic potential of agrivoltaic for

the production of clean energy and industrial cassava in

the Colombian intertropical zone. Environmental

Quality Management, 31(3), 267-281.

J. Mastelic, L. Emery, D. Previdoli, L. Papilloud, F.

Cimmino and S. Genoud, "Energy management in a

public building: A case study co-designing the building

energy management system," 2017 International

Conference on Engineering, Technology and

Innovation (ICE/ITMC), Madeira, Portugal, 2017, pp.

1517-1523.

Laiolo, P., Procopio, R., Delfino, F., Andreotti, A., &

Angrisani, L. (2021, September). The University of

Genoa Savona Campus Sustainability Projects. In 2021

IEEE 6th International Forum on Research and

Technology for Society and Industry (RTSI) (pp. 115-

120). IEEE.

SMARTGREENS 2025 - 14th International Conference on Smart Cities and Green ICT Systems

104

Lazzaretti, A. E., Renaux, D. P. B., Lima, C. R. E.,

Mulinari, B. M., Ancelmo, H. C., Oroski, E., ... &

Santos, R. B. D. (2020). A multi-agent NILM

architecture for event detection and load classification.

Energies, 13(17), 4396.

Le, T. T. H., & Kim, H. (2018). Non-intrusive load

monitoring based on novel transient signal in household

appliances with low sampling rate. Energies, 11(12),

3409.

Leminen, S., Westerlund, M., & Nyström, A.-G. (2012).

Living Labs as Open-Innovation Networks.

Technology Innovation Management Review, 2(9), 6-

11.

Mohseni, S., Khalid, R., & Brent, A. C. (2022). Data-

driven, metaheuristic-based off-grid microgrid capacity

planning optimisation and scenario analysis: Insights

from a case study of Aotea-Great Barrier Island. arXiv

preprint arXiv:2209.10668.

M. Mazidi, R. Khezri, M. Mohiti, L. A. Tuan and D. Steen,

"Effects of Calendar and Cycle Ageing on Battery

Scheduling for Optimal Energy Management: A Case

Study of HSB Living Lab," 2023 IEEE International

Conference on Energy Technologies for Future Grids

(ETFG), Wollongong, Australia, 2023, pp. 1-5.

N. Mohamed, S. Lazarova-Molnar and J. Al-Jaroodi, "CE-

BEMS: A cloud-enabled building energy management

system," 2016 3rd MEC International Conference on

Big Data and Smart City (ICBDSC), Muscat, Oman,

2016, pp. 1-6.

Salazar, G. (2013). Modelos de Mercado, Regulación

Económica y Tarifas del Sector Eléctrico en América

Latina y el Caribe – Colombia. Organizacion

Latinoamericana de Energia (OLADE)

Schuurman, D., De Marez, L., & Ballon, P. (2016). The

Impact of Living Lab Methodology on Open Innovation

Contributions and Outcomes. Technology Innovation

Management Review, 6(1), 7-16.

Silva, M. D., & Liu, Q. (2024). A Review of NILM

Applications with Machine Learning Approaches.

Computers, Materials & Continua, 79(2).

Simões, M. G., & Farret, F. A. (2014). Modeling and

Analysis with Induction Generators. CRC Press.

Zarpellon, B. O. (2024). Sistema integrado para

monitoramento de variáveis energéticas do Instituto de

Ciência e Tecnologia de Sorocaba no contexto de Smart

Campus

Manglar Living Lab: Energy Management Through a Smart Microgrid with Artiﬁcial Intelligence

105