Agent-based modeling for trading wind power with uncertainty in the day-ahead wholesale electricity markets of single-sided auctions

This paper, for the first time, adopts agent-based simulation approach to investigate the bidding optimization of a wind generation company in the deregulated day-ahead electricity wholesale markets, by considering the effect of short-term forecasting accuracy of wind power generation. Two different wind penetration levels (12% and 24%) are investigated and compared. Based on MATPOWER 4.0 software package and the 9-bus 3-generator power system defined by Western System Coordinating Council, the agent-based models are built and run under the uniform price auction rule and locational marginal pricing mechanism. Each generation company could learn from its past experience and improves its day-ahead strategic offers by using Variant Roth–Erev reinforcement learning algorithm. The results clearly demonstrate that improving wind forecasting accuracy helps increase the net earnings of the wind generation company. Also, the wind generation company can further increase its net earnings with the adoption of learning algorithm. Besides, it is verified that increasing wind penetration level within the investigation range can help reduce the market clearing price. Furthermore, it is also demonstrated that agent-based simulation is a viable modeling tool which can provide realistic insights for the complex interactions among different market participants and various market factors.