A Markov Decision Model for Cooperative Virtual Power Plants Market Participation
Résumé
The advent of renewable energy induces new problematic electrical network configurations. Indeed, the renewable energy characteristics (intermittency, production uncertainty) combined with their inherent spatial distribution implies to aggregate them and to design specific control tools to connect them to the grid and allow them to participate to the market. Based on an industrial case study, we propose in this work to gather and coordinate numerous independent producers through a Cooperative Virtual Power Plant (CVPP). To perform an efficient market participation of such coalition of producers, we model its behavior as a Markov Decision Process. Our model takes into account renewable generation prior uncertainty, market constraints and optimizes sequentially the utilization of available resources. Experiments realized using realistic datasets show the efficacy of the proposed model and speak in favor of renewable and controllable producers’ coalition in CVPP.