Artificial Intelligence for Sustainable Energy Transition with Barbara´s Cybersecure Edge Infrastructure

IA4TES Project Ambition

‍

In today’s rapidly evolving energy landscape, accurate predictive methods are essential for managing demand uncertainties and reducing costs. The IA4TES initiative was defined to respond to the challenges of a new energy paradigm characterized by:

• Renewable energy generation from both centralized and distributed systems.
• A digitized, automated grid, delivering bidirectional services to all users.
• Innovative services to enable user participation in new electricity consumption models.
• AI-Enabling Technologies: The exploration of advanced intelligence algorithms and new paradigms for data governance and distributed intelligence.

‍

Advancing Edge Computing in the Power Grid. Barbara's Role in Distributed Intelligence

Edge infrastructure enables real-time data processing at the source, allowing for immediate analysis and decision-making. This reduces latency and ensures rapid responses to dynamic grid conditions like demand spikes or equipment issues, enhancing overall efficiency. By decentralizing operations, Edge Computing improves reliability and resilience.  The following image illustrates the proposed Edge Infrastructure for IA4TES project.

‍

‍

‍

Main challenges in deploying AI in Distributed Environments

Deploying AI models at the Edge is an ambitious journey, starting with data capture, where the diversity of equipment and disorganized input data create significant challenges. Add to this the latency and bandwidth constraints of cloud connectivity, and the need to manage Big Data effectively at the source becomes critical.

Next comes data processing, a stage that tackles the complexity of varied data models and the necessity of data minimization. Establishing common sector-specific ontologies is vital to unify fragmented information, creating a solid foundation for meaningful AI analysis.

With data processed, the focus shifts to its utilization in AI models. Clean, structured data must be ensured, historical databases seamlessly accessed, and real-time requirements met, all while guaranteeing operational continuity even during connectivity disruption

Finally, the journey reaches its critical stage: model retraining. This step ensures that AI systems remain effective and aligned with ever-changing real-world conditions, such as evolving data patterns or shifting operational demands.

To achieve this, advanced frameworks like AIOps (Artificial Intelligence for IT Operations) and MLOps (Machine Learning Operations) play a pivotal role. These frameworks streamline the process of updating, managing, and deploying retrained AI models efficiently, reducing downtime and enhancing performance.

What sets this approach apart is the ability to perform retraining directly at the Edge. By bringing this capability closer to where data is generated, models can adapt in near real-time to local conditions and deliver immediate, context-specific improvements. This localized retraining minimizes latency, reduces reliance on cloud connectivity, and ensures that the AI remains responsive and impactful in the environments that need it most.

‍

Barbara has developed a cutting-edge software that facilitates the remote deployment of AI models at the Edge, enabling the implementation of diverse use cases including optimizing distribution networks and enhancing demand-side flexibility.

‍

AI Orchestration at multiple edge nodes

‍

"I’ve developed an algorithm that works exceptionally well in the lab, but now I need to scale it across 150 locations. How can I approach this effectively?"

Scaling AI is complex, especially when working with diverse environments like substations, transformer substations or distributed energy resources (DERs). To make AI work seamlessly, models must be tailored and trained specifically for each type of asset, ensuring optimal performance across the network.

Once the models are ready, the real challenge begins: scaling and orchestrating AI remotely. That’s where Barbara comes in. Our full-stack technology solution streamlines the deployment and management of AI models, removing the complexities of operational hurdles. It’s a true "plug-and-play" system, designed to simplify the process of deploying and monitoring models seamlessly at the edge.

‍

Curious about the MLOps Workflow at the Edge and how Barbara can support you?  Learn more here

‍

‍

Bringing AI to Life: Real-World Applications in the IA4TES Project

‍

Use Case 1: Enhancing Flexibility to Manage Network Congestion at Transformer Substations

The increasing number of distributed resources across the low-voltage network (the least digitized part of the grid) is leading to a rise in overvoltage and congestion events. The purpose of this use case is to provide insights into how the grid will behave in the future, the identified issues.

The studied infrastructure was designed to respond to:

• Provide continuous data streaming, even when data varies in granularity or access times.

• Create a digital twin of the grid to understand and monitor its real-time status.

• Connect different transformer substations through a Mesh Network of Edge Nodes, enabling visibility into the congestion status and their neighboring nodes.

• Enable MLOps techniques for algorithm execution and retraining.

By combining real-time monitoring, advanced modeling, and AI-driven optimization, this use case demonstrates how to overcome congestion challenges and drive smarter, more resilient grid operations.

‍

‍

Use Case 2: Applying Edge Computing for Flexible Aggregation

This use case highlights how Edge Computing empowers Prosumers to maximize the efficiency of their solar installations by enabling real-time, autonomous decision-making,. These decisions are based on factors like storage availability, energy prices, and production or consumption forecasts, among others.

Prosumers can make smarter energy choices based on key factors such as storage availability, energy prices, and production or consumption forecasts.

‍

‍

From the retailer's perspective, the focus shifts to optimizing the aggregation of diverse Distributed Energy Resources (DERs). Through this approach, retailers can implement innovative solutions like Virtual Power Plants, Energy Communities amongst ohers.

‍

‍

‍

Conclussion

The IA4TES  has set a benchmark for applying AI in the energy sector. In an era of rapid energy system evolution and smart grids, accurate predictive methods  have become crucial for managing demand uncertainties and financial variables. These advancements directly impact sustainability by minimizing energy waste and associated costs. The innovative solutions developed by all members of the consortium have contributed greatly to both, technological progress and environmental sustainability.

Partners

The IA4TES consortium is composed by Iberdrola, Minsait and 9 SMEs including Barbara, 4 Research Centres and 2 Universities. The project has received funding of €12.5 million under the Next Generation EU funds for R&D and Innovation Missions in AI.

‍

‍

‍