Arega Getaneh Abate
Project title: AI-assisted electricity market design for sustainability
Host Institution: Technical University of Denmark (DTU)
Host Supervisor: Dr. Jalal Kazempour
Co-host Institution: École Polytechnique Fédérale de Lausanne (EPFL)
Co-host Supervisor: Dr. Maryam Kamgapour
Summary project: As humans, we cannot fully understand how to model the dynamic interactions of different actors in marketplaces. Most market models are biased and inefficient. Particularly in the energy systems, more challenges are emerging with the ongoing transformation driven by the expansion of renewables sources, demand response, smart homes, and battery storage etc. However, the recent advancement in artificial intelligence enables us to design markets that are both unbiased and efficient.
The objective of this project is to design efficient energy markets considering online learning algorithms to achieve the green energy transition for a sustainable society.
In designing efficient markets, we propose a novel idea by considering three main actors in the power market systems—price-maker agents, demand aggregators and a policy regulator. The price-maker renewable trading agents interact with multiple rivals and learn how to bid efficiently in an adversarial learning environment to improve their regret guarantees. The demand aggregators learn efficient price tariffs to achieve a certain consumption reduction at certain times in a bandit dynamic environment. Regulators learn in repeated games to tune the incentive parameters to reduce CO2 emissions while interacting with generators with unknown decision-making models.
The proposed methods will: (i) create better marketplaces for more participants to enter the market, which increases social welfare, and assist the green energy transition, (ii) increase predictability and provide reliability in the markets, (iii) provide new methods, concepts, and approaches to deal with the energy systems, and to foster future research.