Generation companies' adaptive bidding strategies using finite-state automata in a single-sided electricity market

Abstract

SUMMARY This paper explores the use of genetic algorithms (GAs) in the development of the bidding strategies used by generation companies under two different market clearing mechanisms, uniform pricing and pay-as-bid pricing. The bidding strategies are represented by two modifications of a classical data processing structure known as finite-state automata. Semi-fixed fitness function and co-evolutionary fitness function were incorporated in our GA. A third simple representation to obtain a comparison baseline for the other two representations, showing how their behaviors compare with a “standard” solution, was also incorporated. The strategies developed by our method were adaptive, and all GA types were based on maximizing profit in a competitive bidding situation. Copyright © 2011 John Wiley & Sons, Ltd.