基于多代理决策模型的电力市场月度集中竞价博弈分析  

Game Analysis of Monthly Centralized Competitive Bidding in Electricity Market Based on Multi-agent Decision Model

作　　者：李知远 张翼飞 张政 李天然[1] Li Zhiyuan;Zhang Yifei;Zhang Zheng;Li Tianran(School of Electrical and Automation Engineering,Nanjing Normal University,Nanjing Jiangsu 210042,China;School of Electrical Engineering,Southeast University,Nanjing Jiangsu 210096,China;NARI Technology Co.,Ltd.,Nanjing Jiangsu 211100,China)

机构地区：[1]南京师范大学电气与自动化工程学院,江苏南京210042 [2]东南大学电气工程学院,江苏南京210096 [3]南瑞集团有限公司,江苏南京211100

出　　处：《电气自动化》2022年第5期50-52,56,共4页Electrical Automation

基　　金：智能电网保护和运行控制国家重点实验室开放课题“基于混合仿真的电力市场与碳市场互动仿真分析与调控机制研究”(SGNR0000KJJS1907532)。

摘　　要：当前我国中长期电力市场建设中,对出清机制有着广泛的研讨,却少有定量的分析。采用计算经济学仿真方法,以省级电力市场月度双边集中交易为对象,基于多代理决策模型,对统一出清电价(uniform clearing price,UCP)和按报价支付(pay-as-bid,PAB)两种出清机制进行了仿真分析。定量分析结果表明,基于多代理决策模型可以动态模拟市场交易的博弈过程,得到理论上的收敛价格,在此基础上能够对市场运行效果进行判断。在竞价博弈均衡情况下,两种出清机制对出清电价的影响没有显著差异,但PAB机制增加了市场主体间的无效博弈。同时,分析结果表明供需比的提高可有效提升市场竞争程度。In the current mid-and long-term power market construction in our country,there are extensive discussions on the clearing-out mechanism,but there is little quantitative analysis.Using computational economics simulation methods,with monthly bilateral centralized transactions in the provincial power market as the analytic object,based on the multi-agent decision-making model,simulation analysis of two clearing mechanisms,uniform clearing price(UCP)and pay-as-bid(PAB),was carried out.The quantitative analysis results show that the multi-agent decision-making model can dynamically simulate the game process of market transactions and obtain theoretical convergence prices.On this basis,the market operation effect can be judged.In the equilibrium of the bidding game,there is no significant difference in the impact of the two clearing mechanisms on the clearing price,but the PAB mechanism increases the invalid game between market entities.At the same time,the analysis results show that the increase in the supply-demand ratio can effectively enhance the degree of market competition.

关 键 词：电力市场 集中竞价交易 出清机制 多代理学习算法 寡头市场 竞价策略 

分 类 号：TM73[电气工程—电力系统及自动化] F426.61[经济管理—产业经济]