Agent技术与博弈论方法在发电市场中的应用比较

Comparison of the Applications of Multi-agent Technique and Game Theory to the Bidding Strategies in Power Market

发电厂商竞价上网能否获胜的关键是如何确定合理的报价曲线, 通过建立符合电力市场竞价实际情况的决策方法来获取电厂机组报价曲线又是竞价成功的关键。在运用传统博弈论方法来获取报价曲线面临诸多不太合理的情况下, 文中引入智能 Agent技术, 建立了完全信息下的决策模型, 并用二元进化算法模拟各竞价主体的学习方法,给出了算法模型与学习步骤。通过对智能 Agent技术与传统博弈方法的比较, 并分析其程序运行结果, 表明在电厂机组报价决策中运用智能 Agent技术来研究发电厂的决策报价具有较好的科学性与优越性, 为我国电力市场发电侧投标报价提供了一条新的思路。

For each generating unit, the key to win the bidding is to decide its rational bidding curve. Decision-making support plays an important role in obtaining such bidding curves. Traditional method based on game theory has many defects. Thus this paper applies agent technique in the power market through establishing decision model under complete information conditions, using duality evolution arithmetic to simulate the learning process of the bidders, setting up the arithmetic model and learning method. Case studies and comparison show that the agent technique method outperforms the game theory method.