华南理工大学学报(自然科学版) ›› 2010, Vol. 38 ›› Issue (3): 109-113,122.doi: 10.3969/j.issn.1000-565X.2010.03.019

• 动力与电气工程 • 上一篇    下一篇

智能代理模型在电力市场模拟系统中的运用

王野平1  杨彦1  荆朝霞1  陈皓勇1 陈天恩2   


  1. 1.华南理工大学 电力学院, 广东 广州 510640; 2.西北电网有限公司, 陕西 西安 710048
  • 收稿日期:2009-06-22 修回日期:2009-12-25 出版日期:2010-03-25 发布日期:2010-03-25
  • 通信作者: 杨彦(1983-),男,博士生,主要从事电力市场、电力系统安全与稳定研究.E-mail:yang.yan@mail.suet.edu.cn E-mail:epyanyang@gmail.com
  • 作者简介:王野平(1956-),男,博士生,现就职于国家电力监管委员会,主要从事电力市场研究.
  • 基金资助:

    国家“973”计划项目(2004CB217905);教育部“新世纪优秀人才支持计划”资助项目(NCET080207)

Application of Intelligent Agent-Based Simulation to Electricity Market

Wang Ye-pingYang Yanring Zhao-xiaChen Hao-yongChen Tian-en 2   

  1. 1. Schol of Electric Power, South China University of Technology, Guangzhou 510640, Guangdong, China; 2. Northwest China Grid Company Limited, Xi'an 710048, Shaanxi, China
  • Received:2009-06-22 Revised:2009-12-25 Online:2010-03-25 Published:2010-03-25
  • Contact: 杨彦(1983-),男,博士生,主要从事电力市场、电力系统安全与稳定研究.E-mail:yang.yan@mail.suet.edu.cn E-mail:epyanyang@gmail.com
  • About author:王野平(1956-),男,博士生,现就职于国家电力监管委员会,主要从事电力市场研究.
  • Supported by:

    国家“973”计划项目(2004CB217905);教育部“新世纪优秀人才支持计划”资助项目(NCET080207)

摘要: 基于智能代理的模拟仿真方法已成为电力市场研究的一种新颖而有效的途径.文中结合某一实际区域电力市场模拟系统的构建,介绍了适合模拟发电厂商报价的智能代理学习算法,详细阐述了VRElearning算法、Q-learning算法以及贪婪算法在模拟系统中的运用及实现框架,并分别探讨了学习算法在代理报价收敛问题上的不同处理方式.算例结果表明,智能代理模型及学习算法能够模拟发电厂商的理性竞价行为.

关键词: 电力市场, 智能代理, 仿真, VRE learning算法, Q-learning算法, 贪婪算法

Abstract:

The intelligent agent-based simulation has become a novel and powerful tool in the study of electricity market. This paper deals with the construction of a simulation system of electricity market. In the investigation, first, the intelligent-agent learning algorithms suitable for simulating the strategic bidding of electricity firms are in- troduced. Next, the applications of the VRE-learning algorithm, the Q-learning algorithm and the greedy algorithm to the simulation system are illustrated, and the corresponding implementation frameworks are proposed. Then, the techniques for the learning algotrithms to handle the convergence of agent bidding are discussed. Finally, an exam- pie is performed to test the effectiveness of the intelligent agent-based simulation. The results indicate that the intel- ligent-agent learning algorithms are capable of simulating the rational bidding behavior of electricity firms.

Key words: electricity market, intelligent agent, simulation, VRE-learning algorithm, Q-learning algorithm, greedy algorithm

中图分类号: