Interdisciplinary Institute in Artificial Intelligence

Vincent DEBUSSCHERE,
Associate Professor at Grenoble INP UGA,
vincent.debusschere@grenoble-inp.fr
LinkedIn
Website

DESCRIPTION

The transition to decarbonized energy systems is critical for mitigating climate change. The integration of renewable energy and distributed resources, such as electric vehicles, is driving current changes in energy systems, contributing to their increased complexity and variability. Artificial intelligence offers complementary solutions to optimize energy distribution, storage, and consumption, thereby enhancing system efficiency and reliability. However, the digitalization of energy systems should go hand in hand with their decarbonization and support self-sufficiency. As such, data-driven techniques, the decentralization of intelligence—notably through edge computing—and the interconnection with information and control systems should be developed, prioritizing the human-in-the-loop approach while, above all, ensuring comprehensive, multi-criteria-based improvements to energy systems, minimizing rebound effects, and empowering end-users to take control of their energy consumption. This constitutes the scope of the proposed Chair on Complex Multiscale Energy Systems.

ACTIVITIES

• A new PhD project has just begun to explore how generative artificial intelligence could help monitor vulnerabilities in electrical networks, in collaboration between the G2ELab, the GIPSA-Lab, and the startup Altrans Energy.
• Ongoing discussions should lead to further collaboration between G2Elab and INRIA on data generation for improved distribution grid modeling, planning, and automatic generation, likely in the form of a master’s thesis.
• A master's thesis should focus on the effective implementation of flexibility in a domestic energy system—in this case, a refrigerator—through the creation of a real-world demonstrator at the G2Elab Monitoring and Intelligent Habitat experimental platform.
• An industry-sponsored PhD program will begin at Enedis, focusing on low-cost PV control, edge computing, and optimized onboard capabilities designed for scalability. This PhD program will be jointly supervised by G2Elab and LIG.
• Two postdoctoral fellowships will be established between UGA and the University of Adelaide on decentralized control and modeling of end-user behavior under grid-related constraints (one in France and the other in Australia).

EVENTS

• Roundtable on AI for Smart Grids at the General Assembly of the Smart Grids Institute in Lyon (March 21, 2025).
• Presentation at the Energy- MIAI meeting , in collaboration with Tenerrdis (April 10, 2025).
• Organization of a roundtable discussion with key industry stakeholders on AI and energy during the MIAI days (June 19, 2025).
• Research workshop with KIT to identify areas for collaboration related to energy systems and various technologies, particularly AI (June 23, 2025).
• Chair of a session on “Multi-source, Multi-energy Microgrids” using AI-based techniques at the Symposium de Génie Electrique (SGE) in Toulouse (July 1, 2025).
• Organization of a seminar on AI for energy systems at the G2ELab Scientific Council (in front of approximately 100 researchers from the lab), with Benoit Delinchant and Rémy Rigo-Marinai (July 10, 2025).
• Organization of a special session on grids during the FACET symposium (October 8, 2025, afternoon).
• Presentation of an ongoing research project in collaboration with the University of Adelaide and Monash University aimed at gaining a better understanding of energy consumption through end-users’ responses to simple incentives for flexibility (October 8, 2025, morning).
• On behalf of the CIMES Chair, Grenoble INP ENSE3 hosted the kick-off meeting for a European project aimed at creating a new international master’s program in “Advanced Energy Systems and AI.” Partners include, among others, KU Leuven, UPC Barcelona, KTH, and ESADE (October 9–10, 2025).
• Contributed to the creation of a module on AI for distribution grids in a MOOC on grids and energy systems, to be released by mid-2026, Romain Rombourg.
• Participation in a roundtable discussion with Schneider Electric and the UGA on AI-based solutions for energy systems at the Rencontres Nationales des Entreprises Locales de Distribution (October 16, 2025, afternoon).
   

SELECTED LIST OF PUBLICATIONS

• Model-Free Safe Deep Reinforcement Learning for Grid-to-Vehicle Management Considering Grid Constraints and Transformer Thermal Stress, Z Zhang, R Rigo-Mariani, N Hadjsaid, Y Xu, Engineering Applications of Artificial Intelligence 162, 112529
• A Comparison of Control Strategies for Charging Management in Electric Vehicle Fleets Under Uncertainty, Z. Zhang, R. Rigo-Mariani, N. Hadjsaid, *Energy and AI*, 100522
• Time-Series Clustering: A Benchmark Study on Energy Data with Insights into Demand Response, RK Ahir, B Delinchant, A Easwaran, Engineering Applications of Artificial Intelligence 163, 112892
• "Hybrid Model of Convolutional Autoencoder and Ellipse Characteristics for Unsupervised High-Impedance Fault Detection," J. Yang, B. Delinchant, D. Niyato, N. Hadjsaid, *Electric Power Systems Research* 238, 111166
  
  Conference papers
   
• Reinforcement Learning for the Management of an Electric Vehicle Fleet in a Distribution Grid, Z Zhang, R Rigo-Mariani, N Hadjsaid, 2025 IEEE Kiel PowerTech, 1–6
• Kumar, S. R., Easwaran, A., Delinchant, B., & Rigo-Mariani, R. (June 2024). Real-time Retail Electricity Pricing Using Offline Reinforcement Learning. In Proceedings of the 15th ACM International Conference on Future and Sustainable Energy Systems (pp. 454–458).
• Modeling Local Electricity Consumption by Incorporating Social Media Data Using Natural Language Processing, Y Huang, MS Shahid, A De Moliner, B Delinchant, P Cauchois, IET Conference Proceedings CP922 2025 (14), 263–267
• Ram Kumar, S., Easwaran, A., Delinchant, B., & Rigo-Mariani, R. (June 2025). Improving Demand Response Programs Using Override Signals with Reinforcement Learning. In Proceedings of the 16th ACM International Conference on Future and Sustainable Energy Systems (pp. 603–611).
• Sanchez, F., Mohamed, A., Rigo-Mariani, R., & Debusschere, V. (June 2025). Supervised Learning for the Bidding of Grid-Connected Batteries in the Day-Ahead Market. In 2025 IEEE Kiel PowerTech (pp. 1–7). IEEE.
• Sanchez, F., Debusschere, V., Rigo-Mariani, R., & Labonne, A. (June 2025). Versatile and self-adapting smart home control. In IET Conference Proceedings CP922 (Vol. 2025, No. 14, pp. 308–312).