Renewable Energy Bidding Strategies Using Multiagent Q-Learning in Double-Sided Auctions

Owing to energy liberalization and increasing penetration of renewables, renewable trading among suppliers users has gained much attention created a new market. This article investigates double-auction scheme operated by an aggregator with limited supervision for trading. To ensure beneficial bidding generators end (EUs) considered as agents, multiagent Q-learning (MAQL) based strategy is developed maximize their cumulative reward. Each agent first provides information about supply or demand who will then return the aggregate demand. Without knowing business model aggregator, agents use Q-tables estimate expected reward determine prices accordingly. Finally, coordinates between update on basis amount power bought sold at they bid. The proposed approach can avoid some unnecessary unrealistic assumptions generally made model-based approaches, such assumption knowledge others’ profiles oligopoly; it consider influence strategies market, which cannot be properly addressed conventional proportional allocation mechanism. A numerical analysis using real-world data considering profit maximization shows that outperformed comparable methods in terms profits satisfaction level EUs: iterative double auction approximately 29%, heuristic 39.9%, random 38.1%, NSGA-II-based multiobjective 62%, MOEA/D-based 83.1% average.