Seminario Datalab
Bilevel Real-Time Pricing Model for Smart Grid Using Reinforcement Learning
Ponente: Suzhen Li (University of Shanghai for Science and Technology, PRC)Fecha: martes 18 de noviembre de 2025 - 12:00-12:30Lugar: Aula Gris 1, ICMAT
Resumen:
Reinforcement learning provides a practical approach to real-time pricing in the smart grid, characterized by fluctuating renewable energy output and diverse load demands. This talk introduces the application of reinforcement learning to real-time pricing. A bi-level decision-making framework is proposed, with the power supplier acting as the upper-level agent and consumers as the lower-level agents. Through real-time information exchange between the power supplier and consumers, the system enables the power supplier to minimize cost while allowing the consumers to maximize their welfare. Experimental results demonstrate that the proposed method effectively encourages consumers to adjust their consumption patterns, significantly enhances energy utilization efficiency, and provides essential technical support for the development of future adaptive and intelligent energy systems.
