论文部分内容阅读
在重复博弈的电力市场中,发电商的竞价策略和个体学习行为是个复杂的动态问题。分析发电商竞价策略这一热点问题的最新研究思路和仿真方法。为模拟发电商的自主学习和市场的动态演化过程,引入博弈学习理论,建立基于模仿学习的发电商竞价策略模型,分析不同信息条件下的学习动态过程,及重复竞标时发电商模仿学习对博弈演变及市场均衡的影响。仿真结果表明模仿学习具有较好的收敛性。另外,完全信息条件下,发电商全局学习能引导市场趋于瓦尔拉斯均衡和古诺–纳什均衡二者之间的一个均衡状态;非完全信息条件下,发电商全局学习的结果是市场总体行为收敛到古诺–纳什均衡,而发电商的个体行为则与古诺-纳什均衡状态存在偏差。
In the repeated game of electricity market, bidding strategies and individual learning behavior of power suppliers are complex and dynamic issues. This paper analyzes the latest research ideas and simulation methods of the bidding strategy of power generator. In order to simulate the autonomous learning of power generation companies and the dynamic evolution of the market, the game theory is introduced to establish a bidding strategy model based on imitation learning for generators, and to analyze the dynamic process of learning under different information conditions and the imitation learning of gamers Evolution and the impact of market equilibrium. Simulation results show that imitation learning has better convergence. In addition, under the condition of complete information, global learning by power producers can lead the market to an equilibrium between the Walrasian equilibrium and the Cournot-Nash equilibrium. Under the condition of incomplete information, the global learning result of the power producers is the overall market The behavior converges to the Cournot-Nash equilibrium, while the individual behavior of power producers is in deviation from the Cournot-Nash equilibrium.