Sustainable Smart Electricity Markets

Abstract

High penetration of intermittent renewable energies in power system portfolios induces high electricity procurement costs during peak times to maintain reliability. Ultimately, aggregators pay these costs in the form of capacity charges. This paper proposes an aggregator model as an autonomous trading agent aiming to maximize profit by developing tariffs using offline and online Reinforcement Learning and market segmentation to consumers. The performance of the proposed aggregator is evaluated using a market simulation platform called Power Trading Agent Competition (Power TAC). We find that designing tariffs with usage-based charges and fixed periodic charges can help the aggregator segment the retail market resulting in lower peak demand and capacity charges. CCS CONCEPTS• Computing methodologies → Multi-agent reinforcement learning.