
Project AI-NRGY
Excerpt
Distributed AI architecture for future energy systems integrating a large number of distributed sources
Professor Thierry Monteil, INSA Toulouse – IRIT
AI-NRGY’s objective is to address the major constraints of tomorrow’s energy networks (highly distributed, dynamic, heterogeneous, critical and sometimes volatile) by contributing to the implementation of distributed intelligence solutions taking advantage of different computing methods (edge, fog and cloud computing), and by proposing a software architecture as well as the methods, models and algorithms needed to implement distributed intelligence solutions likely to accelerate the digitization of energy networks.
Keywords: Digitization, Artificial Intelligence, Smart Grids, Distributed AI, Edge 2 cloud
Tasks
Our researches
Use cases
The project will consider different cases in which future energy systems will be used with a very large number of stakeholders, equipment and consequently data sources, and will be intermittently connected.
Software architecture and distributed AI
The focus will be on providing an adaptive distributed software architecture, in terms of service localization and data feedback, capable of satisfying performance, confidentiality or even the characteristics of support equipment and the capabilities of distributed AI algorithms.
Similarly, we plan to develop a distributed AI capable of integrating data distribution, data volatility (e.g. brought about by the use of electric vehicles) and information access dynamics.
End-user focus
The project will strive to improve the interpretability and explainability of AI-based solutions, bring greater robustness to problematic data, and simplify the chain of classical models (forecasting/optimization) in distributed applications to promote their acceptability by end-users and their large-scale deployment in smart grids.
Players will be encouraged to share their data by sharing the added value generated by their use, while respecting privacy and confidentiality.
Consortium
Four institutional partnerships (CEA, INRIA, Mines Paris / PSL, UT3 / IRIT)
A software environment (middleware to the oneM2M standard) to facilitate the digitization of energy systems: deployment on physical devices and simulated equipment will be carried out. A dynamic management of data and services for a distributed digital architecture: this will be used as part of a simulation of an energy network managed by distributed AI. An integrated solution to increase the resilience of distributed learning to cyber-attacks or failures that can compromise distributed AI-based services for smart grids. Distributed AI solutions for collaborative forecasting, optimization and control of energy systems: these will be tested in a simulated environment.
The computing infrastructures required for distributed AI can be major energy consumers. Managing the deployment of AI modules and the localization of the data needed to run them will have to take energy consumption into account.
Training of 7 PhD students and 3 post-docs.

Pas d’actualités
Publications
Les autres projets du PEPR




