TTP2 Winning Projects

Can you describe your project in a few words?
AI4Helios+, a project around the optimization of public lighting systems through AI, coordinated by Connecthink in Spain, is focused on ensuring that public lighting operators can adjust the degree of lighting to a minimum depending on the demand forecast by human activity or weather conditions, using gamification tools to ensure the level of awareness of the impact of decisions. The goal it’s to find a balance between comfort and CO2 emissions. The project will be complemented by predictive maintenance of the public lighting components.

Who will help implement the AI solution?
Connecthink is an SME located in Barcelona with the mission to solve challenges through Artificial Intelligence since 2016. We combine a talented team and deep knowledge of AI technologies. On the other hand, IHMAN, a Malaga (Spain) company, knows the public lighting market, having data and a platform to be used in the project.

What is the AI solution the project plans to implement?
AI4Helios+ wants to improve the public lighting system by adjusting the lighting level according to actual needs. These real needs will be obtained from the weather forecast, moon phase, expected human activity in a specific area and energy price. According to the predictable context, these lighting needs will be proposed through a gamification tool so that the operator is aware of the impact the cost savings and CO2 reduction level that their decisions will have regarding leaving the standard lighting programming. On the other hand, we will incorporate a module for predictive maintenance of the luminaires. The historical data to train the predictive models will be enriched with context information at the weather level that existed at that time.

AIRGo

Can you describe your project in a few words?
The objective of the AIRGo project is to build an open-source end-to-end framework for the Energy Community to develop and evaluate AI-based assistants to flexibly operate power grid for European Power Grid Operators. This platform will leverage existing open-source bricks to bring AI-assisted grid operations closer to control rooms, thanks to the use of industrial components and a validation on real-life data. The ambition is to help human power grid operators avoid blackouts in the near feature, handling fast evolving dynamics at a time of great geopolitical instability and of steep energy transition with large renewable penetration.

Who will help implement the AI solution?
The AIRGo team is composed of two partners, Artelys (lead partner) and RTE. Artelys was founded in 2000 and is a French independent consulting company, specialized in optimization, data science, decision support and modelling thanks to a high-level expertise in advanced quantitative techniques such as statistics, Artificial Intelligence and solutions in optimization. Artelys operates in diversified sectors but is particularly active in the energy sector. RTE (Réseau de Transport d’Electricité) is the French Transmission System Operator (TSO) in charge of the operation, the maintenance and the development of the high and very high voltage grid (100 000km from 63 to 400kV), and also managing the interconnection lines with other European countries. RTE is the largest TSO in Europe and considers AI as one promising avenue for improving grid flexibility in real-time operations.

What is the AI solution the project plans to implement?
The AI solution to implement will rely on two existing frameworks (Grid2Operate and PowSyBl). Grid2Operate is an open-source python framework that allows to perform powergrid operations. It is based on the OpenAI Gym standard for sequential decision-making problems with AI and is listed as a reference framework for instance here. It comes with a code Interface for connecting any physical simulator in the backend. Open-source but non-industrial physical simulators have been connected so far. On the other end, PowSyBl is an open-source industrial physical simulation framework deployed for real-time operations today in Europe. It is able to handle real-grid format and data. Implementing a bridge between PowSyBl and Grid2Op (in the Grid2Op backend standard) will make the overall framework ready for use on real-grids.

E-ModelOps

Can you describe your project in a few words?
Rebase.energy is a digital energy startup based in Stockholm, Sweden, closely connected to the KTH (Royal Institute of Technology) ecosystem. Our vision is to accelerate the energy transition by empowering energy innovators with data and digital tools.
In the I-NERGY project, we are building the world's first open platform for energy modelling. The platform targets energy engineers and data scientists to enable them to create better energy models to forecast and optimise distributed energy systems. Two different AI models focused on wind power forecasting will be developed in the project.

Who will help implement the AI solution?
The Rebase Team will develop the AI models with the project partner Modity Energy Trading. Close collaboration with the users in the development process will ensure the usefulness and business value of the developed AI models.

What is the AI solution the project plans to implement?
The AI solution is based on two complementary solutions. Firstly, a vector autoregressive approach for short-term wind power forecasting and a multi-model blending approach to improving day-ahead-horizon wind power forecasting.

EMERGENT

Can you describe your project in a few words?
The EMERGENT project aims to develop and demonstrate a user-driven Energy Management System (EMS) for energy efficiency in facilities. Specifically, EMERGENT leverages the available historical and real-time data related to a facility. It merges them with consumer-level energy demand forecasting and user feedback data to offer an EMS solution with higher levels of user acceptability and hence energy efficiency for facilities and buildings.

Who will help implement the AI solution?
Plegma Labs will implement the AI solution. Plegma Labs bridges protocol barriers and applies meaningful rules and workflows that add intelligence to its applications, leading to efficiency & optimization. Plegma Labs has substantial experience in AI & data analytics applications in energy, such as predictive maintenance, Photovoltaic (PV) generation forecasting, and energy optimization. Plegma's IoT platform product also includes data analytics services for Internet of Things (IoT) data, such as demand forecasting (https://pleg.ma/platform/) "smartCard-inline".

What is the AI solution the project plans to implement?
EMERGENT utilizes Reinforcement Learning (RL) techniques to train an agent that provides energy efficiency recommendations for specific facility assets, e.g. Heating, Ventilation and Air Conditioning (HVAC) and boilers. In addition, Deep Learning (DL) models are used to conduct high-granularity building energy demand forecasts, even in cases of limited historical data availability. The EMERGENT system continuously considers user feedback regarding energy efficiency recommendations to ensure the RL agent learns the end-user behaviour. The idea is to constantly train the RL agent for each facility with historical and real-time energy and non-energy data for the available assets.
The agent's goal is to produce asset/device re-scheduling recommendations for facility/building managers to minimize energy costs while considering explicit and implicit user feedback and preferences. The agent also finds high-granularity building-level energy demand forecasts to enhance its performance further.

SPIRE

Can you describe your project in a few words?
SPIRE, led by Plaixus, a deep-tech company from Greece, will produce an Edge-AI platform, built on top of open-source technologies such as ROS and KubeEdge, to be offered as an add-on to robotic solutions enabling predictive maintenance of Photovoltaic (PV) panels.

Who will help implement the AI solution?
SPIRE is being developed in collaboration with ELVAN, a leading manufacturer of PV Panel structures, with over 100 sizeable solar farm operators among their clientele, who are spinning out a new company, Solbotix, which will commercialise a robotic PV panel cleaner. ELVAN provides access to their operational PV panels and will facilitate engagement with solar farm operators to validate the complete SPIRE platform.

What is the AI solution the project plans to implement?
SPIRE is an Edge-AI Platform that will facilitate AI service providers to develop, train and deploy federated models at the edge, accelerating the adoption of solar energy AI applications. It will be implemented as an add-on to cleaning robotic solutions. It will enable solar farm operators to streamline their maintenance schedules, facilitating rapid deployment of data-driven, AI-powered predictive maintenance algorithms. By the end of the project, SPIRE will comprise ML models trained on both images and vibration signals to detect dirt and loosened bolts, respectively.

GEM-GREEN

Can you describe your project in a few words?
GEM-GREEN aims at developing new AI-based gamification analytics and a mobile app capable of interacting with personnel by suggesting a list of daily activities (engagement) to improve buildings' energy efficiency and well-being for customers/employees through gamification mechanisms (rewards, competition with other shops/factories).

Who will help implement the AI solution?
The solution will be implemented by Energenius Srl, an innovative small-scale Italian start-up owned by Maps SpA, that integrates existing hardware products for energy monitoring systems and offers added value through advanced energy consumptions analysis, carried out directly by its experts and its advanced software solutions based on AI, in collaboration with Calzedonia SpA, one of the biggest retail company in underwear, bathing suits, tights, and leggings, with over 5,000 shops worldwide.

What is the AI solution the project plans to implement?
The GEM-GREEN project aims to study, test and eventually adopt a series of AI-based gamification tools and a brand-new mobile app. In the energy monitoring sector, no solution currently combines energy analysis and gamification strategies. Our ambition is to include off-the-shelf solutions for retail businesses which will positively influence the behaviour of the employees towards efficient consumption using a gamification approach. For this reason, GEM-GREEN will not only improve the energy efficiency of the buildings but will also have a strong positive effect on human behaviour by leveraging AI and data-driven decisions.

ARIA

Can you describe your project in a few words?
ARIA (enhAnced shoRt-term wInd forecAsts) is an AI-based tool for providing improved short-term wind forecasts to the wind energy sector, enabling better predictability of wind power availability and generation. ARIA is developed by Amigo srl (as project coordinator and technology service provider) and ANEV (the Italian Association for Wind Energy, EPES stakeholder and Pilot infrastructure provider/Data Owner entity).

Who will help implement the AI solution?
The ARIA ARIA consortium comprises Amigo srl (as project coordinator and technology service provider) and ANEV (the Italian Association for Wind Energy, as EPES stakeholder and Pilot infrastructure provider/Data Owner entity). Amigo is the first Italian SME working on climate services. We leverage climate data to deliver personalized solutions for clients to assess, manage, and forecast climate-related risks in several sectors, such as energy, insurance, infrastructure, and agriculture. Amigo is a multidisciplinary team of professional figures from climate science to Big Data and Artificial Intelligence, from strategic design to business development. Business experts support physicists and computer scientists to develop services that address specific customer needs.
ANEV is the Italian Association for Wind Energy, comprising approximately 100 companies involved in the national wind energy sector. It represents various stakeholders in the wind energy value chain, such as electricity producers and operators, technology providers, engineering firms, environmental studies, electrical traders, and developers.

What is the AI solution the project plans to implement?
ARIA aims to overcome the limited accuracy of the meteorological models already in use in predicting wind speed and direction by enhancing their output data through AI. By leveraging a set of AI and ML-based approaches (e.g., time series forecasting, statistical downscaling) on the zonal and meridional components of wind, ARIA allows the forecasting of wind speed at locations of interest up to 24-72h in advance.

FlexyGrid

Can you describe your project in a few words?
FlexyGrid contributes to renewable energy management and decentralised energy transition by implementing an Artificial Intelligence (AI) powered solution to forecast and optimise Renewable Energy Communities (RECs). It predicts renewable energy production and consumption patterns, focusing on high-frequency, low-latency micro-site predictions for multiple energy sources such as solar and wind power. These predictions can facilitate effective energy management, enabling communities to optimally balance internal and external energy flows, share energy efficiently, and contribute to a greener and more sustainable energy landscape.

Who will help implement the AI solution?
The AI solution is implemented by a highly specialised and dedicated team, which comprises PhDs, MBAs, talented data scientists, machine learning experts, software developers, and energy domain specialists. Furthermore, we engage with various stakeholders, including, but not limited to, local Renewable Energy Communities, technology providers, energy management system vendors, and regulatory bodies. Their insights, coupled with our team's expertise, ensure our AI solution's successful implementation, integration, and widespread adoption.

What is the AI solution the project plans to implement?
The FlexyGrid AI solution comprises state-of-the-art deep learning, conventional statistical modelling, and advanced reinforcement learning methodologies. We harness the power of these approaches to create a hybrid forecasting model capable of capturing the complex and variable nature of renewable energy patterns.
Our system ingests data from diverse sources like weather stations, IoT devices, smart meters, and user-generated inputs. Following data cleaning and pre-processing, the engine performs feature engineering to extract meaningful insights and normalise the data, as well as learning techniques to combine individual forecasting models, generating more accurate and robust predictions. Our solution offers a RESTful API and an intuitive user interface for end-users to ensure smooth integration with existing energy management systems. By leveraging cutting-edge containerisation technologies such as Docker and Kubernetes, we provide our solution can be deployed at scale, efficiently and reliably.

CNROpt

Can you describe your project in a few words?
CNROpt is an ambitious project to develop an innovative tool for power companies and operators of Charging Station Networks for Electric Vehicles. It will help facilitate the efficient rollout of new Charging Station Locations and the expansion of the existing ones by suggesting the optimal sites for growth, considering both business and customer experience requirements. Powered by AI, we will develop new AI Algorithms based on recent research in the field of AI and train the models with a rich set of time series data.
These data are the results of anonymous charging session raw data of several hundred of thousand sessions from more than 250 locations in Croatia.

Who will help implement the AI solution?
SLOA LTD is a technology SME founded in Cyprus that has established a branch in Kalamata, Greece, in 2022 for AI R&D, branded as Local AI, developing innovative solutions for sustainability utilizing the power of AI for local and global stakeholders, such as Municipalities, Regional Government, Private Energy Stakeholders, etc.
AI innovation is at the core of its business strategy, and it actively participates in research and innovation projects to jump-start the development of new functionalities and commercial solutions. We envision a future in which AI is ubiquitous and assists local, regional, and global green transition initiatives. We work with the University of Zagreb, with their deep domain knowledge of the EV charging network technologies, to provide a disruptive way to develop Charging Networks in Europe.

What is the AI solution the project plans to implement?
For the Timeseries Forecasting models, we plan to try three algorithms:

For the Candidate Stations Scoring – Selection modelling, we will approach this problem by trying to geographically cluster the existing Charging Stations with the K-means Clustering in Geographical Dimensions based on various KPIs (RoI, Utilization, Congestion, waiting times, etc.) as the previous step produces them, the Timeseries Forecasting. This will help us detect the geographical areas indicating hot spots for the Charging Station Network, prioritizing the relevant candidate sites for expansion.

Prophet off-the-self algorithm, a widely used procedure for forecasting time series data based on an additive model where non-linear trends fit yearly, weekly, and daily seasonality, plus holiday effects. It works best with time series with substantial seasonal effects and several seasons of historical data. Prophet is robust to missing data and shifts in the trend and typically handles outliers well. The Prophet procedure includes many possibilities for users to tweak and adjust forecasts. By adding your domain knowledge, we will use human-interpretable parameters to improve our projections.
Transformer Architecture for Timeseries. Transformers constitute a class of Deep Learning Models suited to handle temporal dependencies on data. They are the backbone of modern Language Models (GPT. We will develop our custom Transformer-based prediction model to perform the congestion-utilization prediction.
LSTM. Long short-term memory (LSTM) is an artificial neural network in artificial intelligence and deep learning. Unlike standard feedforward neural networks, LSTM has feedback connections. Such a recurrent neural network (RNN) can process not only single datapoints (such as images) but also entire sequences of data (such as speech or video). This characteristic makes LSTM networks ideal for processing and predicting data.

THRUST-AID

Can you describe your project in a few words?
THRUST-AID's long-term goal is to develop an integrated drone + AI system for electricity grid diagnostics, which will contribute to reducing economic losses, inspection costs, and environmental impact by providing a reliable AI-based tool for timely automatic defect detection in aerial imagery of the transmission power grid.

Who will help implement the AI solution?
THRUST-AID brings together a dedicated, balanced, cross-functional team with combined decades of experience. Our lead AI developer Karolis has 4+ years of experience developing AI-based solutions for Lithuania's leading IT companies. Indrė (Chief Innovation Officer, PhD in applied physics) has led R&D of custom analytics solutions for multi-sensor UAV data since 2019. Irmantas (an expert engineer with 17+ years of experience in grid maintenance as well as a master's degree in electrical technology and management) will be representing the client's needs and consulting on the technical aspects of inspections.

What is the AI solution the project plans to implement?
THRUST-AID is based on deep-learning models for computer vision that are finetuned using annotated ultra-high resolution aerial imagery. During this project, we will use the extensive, high-resolution, real-life data of the Lithuanian electricity transmission grid to train automated AI/ML-based image recognition algorithms to identify the most common electricity grid elements and their defects. Each detected defect will have assigned a unique ID based on GIS metadata and classified depending on severity & risk and reported automatically via a user-friendly interface to enable predictive maintenance.

SuperPower v2.0

Can you describe your project in a few words?
The goal of the SuperPower v2.0 is to develop the first fully automated power line inspection value chain powered by superresolution. Currently, over 100 million kilometres of power lines are being inspected mainly using crewed helicopters, which are inefficient, pollutant and have personal risk.
Drones could perform this kind of inspection, but one of the main challenges is the quality of the images taken by drones, as the sensors onboard the drone are usually much smaller and with lower resolution than crewed helicopters’ sensors. That is why we are working to integrate superresolution technologies to enhance the quality of the imagery taken by drones automatically.
The project will build on top of the previous project (SuperPower V1.0), where we achieve to develop the first superresolution system to enhance visual photos for power line inspections. In this second project, we are developing the following:

A superresolution module to enhance thermal images.
Integrating both superresolution modules (visual + thermal) in the inspection process.
Creating an automatic power line defect detection with this imagery.

Who will help implement the AI solution?
We represent the whole value chain of the solution: FuVeX as a drone developer and operator, ATLAS as a data analysis company, and UFD-Naturgy as the third most significant Spanish utility. Consequently, we will work on a TRL7 solution that verifies the product-market fit. The goal is to continue the development after the project has ended, achieving TRL9 in 2025.

What is the AI solution the project plans to implement?
We aim to deploy superresolution technologies to enable the use of drones in power line inspection, achieving the same data quality as crewed helicopters.

GRIDouble

Can you describe your project in a few words?
GRIDouble is a comprehensive energy management tool that completely automates finding optimal patterns in energy consumption and production in the case of facilities with renewable energy sources. It will be offered to end users in the form of SaaS, thus enabling them to optimize energy savings without investment in scarce available data science and optimization experts or the necessary computing infrastructure.

Who will help implement the AI solution?
Vodena is a company focused on R&D activities with its primary interest in software development in physical and data-driven modelling and simulation and optimization of both social and natural phenomena. We have developed several software solutions, including general data management, hydrological modelling and simulations, power production optimization, and data visualization, operational in the hydro-power industry. Our most outstanding achievement so far is developing and deploying a comprehensive data analysis and optimization solution employed at the “Iron Gate” hydro-power plant on the Danube River. In the previous years, we have successfully participated in several EU projects focused on implementing cutting-edge technologies in the energy sector and other industries.

What is the AI solution the project plans to implement?
Our innovative approach uses automated machine learning (AutoML) to automatically generate data-driven predictive models of diverse energy sources, consumers, and storage within a system. Based on the data acquired during the grid exploitation, GRIDouble automatically creates adequate predictive models of energy production and consumption, which are improved by using publicly available data on weather, working days, electricity prices, solar irradiation, etc.

AI4CZC

Can you describe your project in a few words?
AI4CZC aims to build a web application to provide accurate key elements allowing Cross Zonal Capacity (CZC) estimations for Transmission System Operators (TSOs) in the European energy market. During this project, we will build models to help the Montenegro TSO to estimate CZC 48h in advance.

Who will help implement the AI solution?
The project will be led by Inceptive, a French SME specialising in developing Machine Learning solutions, and CGES, the Montenegro TSO.

What is the AI solution the project plans to implement?
During AI4CZC, we plan to build an AI platform using Inceptive's Machine Learning platform, Igloo. This platform will automatically retrieve data from valuable sources, process it and generate predictions using supervised learning models. Then the platform will provide interfaces to access the live forecasts and exploit the models. An API will allow the integration of the models on TSO Information Systems, and a GUI, including a dashboard, will provide human-readable interfaces to assert the correctness of the forecasts.

AI Flex in Edge

Can you describe your project in a few words?
The main goal of this project is to make accurate forecasts on the network’s future behaviour, taking into account, among other topics, problems related to the increasing integration of Distributed Energy Resources in LV networks.
For Medium and Low voltage networks, with thousands of electrical Transformation Centres (from now on, TC) widely distributed geographically, with coverage and connectivity problems, and with near-real-time needs in taking power decisions, Cloud is an option with limited viability due to latency, cost or scalability problems. SCADA technologies, highly oriented towards automation and with proprietary data structures, are not flexible and accessible and do not optimally cover the needs.
The solution is halfway between the Cloud and SCADAs: it is called Edge Computing. The consortium members intend to distribute data analysis and decision-making among the different Edge Computing nodes, or distributed computing, located in adjacent TCs (e.g., in the same Medium Voltage ring) instead of sending this data to centralised servers (Cloud).

Who will help implement the AI solution?
At Barbara, we are fast-tracking the digitalisation of the industry. Our mission is to facilitate the deployment of Artificial Intelligence at the Edge. We are the Challengers sector with a clear vision of where the future is headed and with the technological solution that allows for the creation, deployment and execution of Edge Computing algorithms in a secure and scalable way.
Cuerva is the partner who will help implement the AI solution. Cuerva is a multidisciplinary company within the energy sector whose main activities are Distribution, Generation, Commercialisation and R&D&I activities.
Cuerva is leading the digitisation of its network and utilising cutting-edge techniques to control and monitor its components. Thanks to its previous efforts, Cuerva has achieved the advanced digitisation of some strategic points in its network. One of those areas is the one proposed in this project, Lachar (Granada), whose network is highly digitised and monitored by implementing the Barbara IoT technology in the Edge Node. With these systems already installed, we propose their improvement:

Including new Cuerva algorithms will allow the DSO to fully understand the network’s behaviour and predict future problems in the short term.
Proposing practical solutions to those issues via the actions of various Flexibility Service Providers (FSPs).
Integration of these algorithms and decisions into a replicable framework will allow other European DSOs to improve their network’s state.

What is the AI solution the project plans to implement?
The objective of the AI Solution is to develop a forecasting algorithm that will be able to overcome the limitations described before and predict, with accuracy, the demand and production values of the consumers connected to the transformation centre. During this project, a Digital Twin of the grid will also be developed, and through this, Cuerva will be able to detect events in the grid that could put the supply at risk.
The increasing number of Distributed Energy Resources throughout the Low Voltage Grid (a less digitised part of the grid itself) is increasing overvoltage and congestion events. This project aims to have a perspective on how the grid will behave in the future and to address techniques that solve the problems detected.

AI4Hydro II

Can you describe your project in a few words?
LexaTexer provides an Enterprise AI platform to support the energy value chain with prebuilt, configurable AI applications addressing CAPEX-intense hydro assets like Francis and Pelton turbines and pumps.

Who will help implement the AI solution?
Hydropower operators face several challenges due to the introduction of stochastic energy providers like wind and solar. They are switching from baseload to more flexible power production, introducing a stochastic usage pattern. At the same time, availability and efficiency must be improved. Static wear models no longer suffice; intelligent AI-driven condition monitoring and diagnostics promise remedy.

What is the AI solution the project plans to implement?
In this project, we propose combining our AI platform and data from real-world operations to model Pelton turbines' remaining useful life (RUL) based on real-world operational and environmental data to increase RUL's efficiency and availability significantly. We propose to build hybrid AI models, including operational knowledge into the models.

I-NERGY 2nd Open Call Winning Projects

Announcing the winners of I-NERGY's Second Call: Get Acquainted!

Be the first to get our news

Stay in touch