EnergyGuard

EnergyGuard aims to develop, kickstart and sustain an open, green and robust Testing Experimentation Facility (TEF) operating under real-world conditions to empower innovators in bringing trustworthy AI products to the energy market in a cost-effective manner. 

The project integrates five significant European large-scale testing and experimentation facilities that cover the full energy value chain, supported by European’s greenest HPC infrastructure (Meluxina). This includes a digital twin (DT) of the Portuguese Transmission Network (RDN), the CEDER-CIEMAT Microgrid with its Distributed Energy Resources (DERs), the Hydrogen testing platforms at CEA LITEN, CARTIF, BER and CIEMAT, a high-fidelity local DT of Riga's multi-apartment residential buildings and the Antrodoco Renewable Energy Community. The process requires a wide range of elements to cover diverse AI test needs, including wind power, photovoltaic systems, hydropower plant, AEM, PEM and SO eletrolyzers, fuel cells, EV charging stations, electric and public buses and battery storage systems. The facilities will be accessible to EnergyGuard end-users through a set of properly configured Digital Twins (DTs) and curated assets, including data, models, inference APIs, services, and applications through an AI development Testing environment. 

EnergyGuard enables seamless access to assets from the EU ecosystem including AIOD, Data Spaces, DIHs and other TEFs. Moreover, the project facilitates users to validate their products within an Acceptance Environment and a common open AI risks database against a wide range of cybersecurity and trustworthy AI assessments. 

The TEF will serve as full infrastructure to support national AI regulatory sandbox initiatives and deliver 5 pilot cases for the private and public sector. EnergyGuard will build upon a long-term, self-sustainable business model driven by a new entity, incorporating market-ready features early in the design, such as a subscription/plan framework, billing, and professional support. 

Project Objectives:

In a nutshell, the objectives of EnergyGuard comprise the following:

• Design, develop and deploy an open, green, robust and scalable TEF covering the entire energy value chain, operating under real-world conditions to empower innovators to bring trustworthy AI solutions to the energy market cost-effectively.
• Deliver a seamlessly integrated framework of tools and services to enable AI innovators to design, build, execute and monitor AI pipelines, improving innovation, flexibility, cybersecurity, resilience and environmental impact of energy systems.
• Establish and pilot a dynamic environment for assessing trustworthiness and ensuring quality assurance of AI solutions in the EPES sector, supporting AI regulatory sandboxes.
• Create a publicly disclosed database of ethical and environmental risks for the energy sector, under the supervision of national authorities.
• Strengthen European-wide R&I in AI by liaising with and extending relevant initiatives (AI-on-Demand, DOME, EDIHs, TEFs from other sectors, robotics platforms, EU data spaces).
• Initiate, demonstrate and assess EnergyGuard TEF services through experiments and AI solutions in five impactful use cases.
• Promote international collaboration and adoption of the project’s outcomes among AI innovators (especially SMEs and start-ups) and stakeholders in AI, HPC, Big Data, Security, and Ethics.
• Establish commercial features and a business plan to guarantee exploitation, self-sustainability and growth of the EnergyGuard TEF beyond the project’s lifetime.

ICCS role:
ICCS (Decision Support Systems Laboratory, NTUA) is the coordinator of EnergyGuard. It represents the consortium in front of the European Commission and manages the project’s scientific, technical, and administrative implementation. Additionally, it is leading activities such as the architectural design and interoperability of the TEF, the semantic data model, AI experimentation pipelines and user services, as well as the AI trustworthiness assessment framework and common AI risk database.