Learning models to simulate electricity markets

In this paper, we explore the reasons for electricity price diversity in electricity market from the view point of game learning theory. It is assumed each supplier can choose a learning model to help him to make strategy selection from Cournot adjustment and fictitious play learning models, which can produce two single learning models (when all suppliers choose the same learning model) and numerous multiple learning models (when different supplier chooses different learning model).Simulation results show numerical multiple learning models bring large price diversity to electricity market which can lead to market instability. Therefore, market designers should develop market rules to avoid this kind of behavior risks.