容量电价下电-碳市场多智能体两阶段市场操纵行为仿真

杨宣 , 苏珈 , 杜欣慧 , 梁淦 , 钱帅帅 , 于昊扬

南方电网技术 ›› 2026, Vol. 20 ›› Issue (6) : 46 -57.

PDF (3654KB)
南方电网技术 ›› 2026, Vol. 20 ›› Issue (6) : 46 -57. DOI: 10.13648/j.cnki.issn1674-0629.2026.06.004

容量电价下电-碳市场多智能体两阶段市场操纵行为仿真

作者信息 +

Simulation of Multi-Agent Two-Stage Market Manipulation Behavior in the Electricity-Carbon Market Under Capacity Pricing

Author information +
文章历史 +
PDF (3741K)

摘要

在新型电力系统的背景下,传统发电厂商或许会借助市场机制行使市场力,以获取额外收益。为防范发电商可能采取的各类市场操纵策略,从监管视角提出了一种综合考量多种市场行为的策略仿真模型,用以模拟发电商的行为。首先,剖析了碳背景下多侧市场的耦合关系,并依据市场机制分析了可能影响市场出清的因素。其次,鉴于单一行为存在的不足,全面考虑发电商可能实施的多种操纵手段,涵盖容量、价格申报以及串谋行为,构建双层竞价模型,分别为发电商决策层和市场出清层。然后,鉴于决策变量包含离散和连续两种类型,采用DDQN和TD3算法相结合的方式分别求解串谋意向和信息申报的两阶段动作决策问题,模拟各发电商策略的演化进程。仿真结果表明,在多侧耦合的市场环境中,发电商会倾向于与具备竞价优势的个体强强联合,实现暗中串谋,形成联盟并操纵电价及容量,从而降低市场效率。同时还发现,合理的市场机制能够在一定程度上减少容量持留现象,缓解市场力,可为市场监管提供一定的参考价值。

Abstract

Under the backdrop of the new power system, traditional generation companies may leverage market mechanisms to exercise market power and gain additional profits. To prevent potential market manipulation strategies by generators, from a regulatory perspective, a comprehensive strategy simulation model considering multiple market behaviors is proposed to simulate the behavior of power generators.. Firstly, the coupling relationships of multi-side markets in the carbon context are analyzed, and factors potentially influencing market clearing are examined based on market mechanisms. Secondly, recognizing the limitations of single behaviors, the model holistically considers various manipulation which tactics generators might employ, including capacity withholding, price bidding, and collusion. A dual-layer bidding model is constructed, consisting of the generator decision-making layer and the market clearing layer.Then, given that decision variables include both discrete and continuous types, a hybrid approach combining DDQN and TD3 algorithms is adopted to solve the two-stage action decision problem involving collusion intentions and information bidding, simulating the evolution of each generator′s strategy. Simulation results reveal that in a multi-side coupled market environment, generators tend to form strong alliances with competitors possessing bidding advantages, enabling tacit collusion to manipulate electricity prices and capacity, thereby reducing market efficiency. Additionally, it is found that reasonable market mechanisms can partially mitigate capacity withholding and alleviate market power, offering valuable insights for market regulation.

Graphical abstract

关键词

竞价策略 / 多智能体 / 容量电价 / 深度强化学习 / 串谋行为

Key words

bidding strategy / multi-agent / capacity pricing / deep reinforcement learning / collusive behavior

Author summay

杨宣(1999),男,硕士研究生,研究方向为电力市场、人工智能,

苏珈(1995),女,副教授,博士,研究方向为电力系统稳定与控制、综合能源系统,

引用本文

引用格式 ▾
杨宣,苏珈,杜欣慧,梁淦,钱帅帅,于昊扬. 容量电价下电-碳市场多智能体两阶段市场操纵行为仿真[J]. 南方电网技术, 2026, 20(6): 46-57 DOI:10.13648/j.cnki.issn1674-0629.2026.06.004

登录浏览全文

4963

注册一个新账户 忘记密码

参考文献

基金资助

国家自然科学基金资助项目(52307130)

山西省青年科学基金资助项目(202303021212058)

AI Summary AI Mindmap
PDF (3654KB)

57

访问

0

被引

详细

导航
相关文章

AI思维导图

/