Stratégie optimisée de contrôle de statisme assistées par réseaux de neurones pour la gestion des convertisseurs de puissance monophasés dans les micro-réseaux intelligents

Abstract

In the context of current energy and environmental challenges, the strategic importance of microgrids, particularly for the integration of decentralized energy resources, is undeniable. These systems are characterized by their ability to integrate various renewable energy sources and ensure efficient electrification of isolated sites, thus constituting a viable alternative to centralized energy systems. However, the integration of renewable energy sources, notoriously intermittent, poses significant challenges in terms of performance, stability, and power management. This thesis addresses these challenges, focusing particularly on the management of single-phase power converters within microgrids. A vector control method is proposed, which has demonstrated significant improvement in robustness and reliability, addressing the complexities specific to microgrids. In autonomous mode, a droop control system based on Artificial Neural Networks is developed for balanced power sharing between distributed generators operating in parallel. This system offers adaptive and precise power regulation, particularly responsive to the variabilities of renewable sources. In grid-connected mode, a droop control method, also based on Artificial Neural Networks and compliant with IEEE1547 standard, is introduced. This approach ensures harmonious integration of distributed generators with the main electrical grid, while respecting the standard’s rigorous criteria for stability and performance. For both modes, the Artificial Neural Networks were trained using data from detailed simulations, and the operating points were optimized through a particle swarm optimization method. This strategy enables optimal management of microgrids, conferring upon them the ability to dynamically adapt to the volatility of renewable energy sources. The obtained results demonstrate that the proposed control strategy significantly improves stability and power distribution management in microgrids, both in islanded and grid-connected modes. This advancement is crucial for the development of more efficient and effective distributed power systems, facilitating more stable and efficient integration of renewable energies into future energy distribution systems.

Date	18 Dec 2025
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Handy Fortin Blanchette (Supervisor)