Research and application of the Q-learning for wholesale power markets

The open competitive power market makes it possible for participants to gain more profits through strategic bidding. This paper proposes a model being able to the generators compared with the rivals through supply offers and the dispatch load units are able to change their demand bids. Reinforcement Q-learning is considered and it is used to achieve the goal of improving profit for generators. The model is test for 6-bus transmission grid Wood and Wollenberg. The electricity prices are determined at each bus, physical location of the market participants. The experimental results are used to demonstrate the feasibility of the algorithm suitable with the model.