Auction Design through Multi-Agent Learning in Peer-to-Peer Energy Trading

Abstract: Distributed energy resources (DERs), such as rooftop solar panels, are growing rapidly and are reshaping power systems. To promote DERs, feed-in-tariff (FIT) is usually adopted by utilities to pay DER owners certain fixed rates for supplying energy to the grid. An alternative to FIT is a market-based approach; that is, consumers and DER owners trade energy in an auction-based peer-to-peer (P2P) market, and the rates are determined based on supply and demand. However, the auction complexity and market participants' bounded rationality may invalidate many well-established theories on auction design and hinder market development. To address the challenges, we propose an automated bidding framework based on multi-agent, multi-armed bandit learning for repeated auctions, which aims to minimize each bidder's cumulative regret. Numerical results indicate convergence of such a multi-agent learning game to a steady-state. Being particularly interested in auction designs, we have applied the framework to four different implementations of repeated double-side auctions to compare their market outcomes. While it is difficult to pick a clear winner, $k$-double auction (a variant of uniform pricing auction) and McAfee auction (a variant of Vickrey double-auction) appear to perform well in general, with their respective strengths and weaknesses.