Aktivitäten pro Jahr
The rise of distributed energy generation technologies along with grid constraints, and conventional non-consumer centric business models, is leading many to explore alternative configurations of the energy system. Particularly popular are peer-to-peer energy trading models in which the role of the energy company is replaced with a trustless transaction layer based on a public blockchain. However, to ensure stable operation of microgrids, an energy company is required to constantly balance supply and demand. In this paper, we study the problem that arises from the conflicting goals of prosumers (to make money) and network operators (to keep the network stable) that have to co-exist in future energy systems. We show that prosumers can play large-scale synchronization games to benefit from the system. If they synchronize their actions to artificially increase energy demand on the market, the resulting power peaks will force the microgrid operator to use backup generation capacities and, as a consequence, contribute to the increased profit margins for prosumers. We study synchronization games from a game-theoretical point of view and argue that even non-cooperative selfish prosumers can learn to play synchronization games independently and enforce undesired outcomes for consumers and the grid. We build a simple model where prosumers independently run Q-learning algorithms to learn their most profitable strategies and show that synchronization games constitute a Nash equilibrium. We discuss implications of our findings and argue the necessity of appropriate mechanism design for stable microgrid operation.
|Publikationsstatus||Veröffentlicht - 31 Okt 2018|
|Veranstaltung||IEEE SmartGridComm: International Conference on Communications, Control, and Computing Technologies for Smart Grids: Workshop on AI in Energy Systems - Comwell Hotel Hvide Hus, Aalborg, Dänemark|
Dauer: 29 Okt 2018 → 31 Okt 2018
|Konferenz||IEEE SmartGridComm: International Conference on Communications, Control, and Computing Technologies for Smart Grids|
|Zeitraum||29/10/18 → 31/10/18|