协同粒子群算法在电力市场ACE仿真中的应用.pdfVIP

第34 卷 第2 期 电 网 技 术 Vol. 34 No. 2 2010 年2 月 Power System Technology Feb. 2010 文章编号：1000-3673 （2010 ）02-0138-05 中图分类号：TM 743 文献标志码：A 学科代码：470·4054 协同粒子群算法在电力市场 ACE 仿真中的应用 陈乃仕，王海宁，周海明，李伟刚 （中国电力科学研究院，北京市海淀区 100192 ） Application of Cooperative Particle Swarm Algorithm in Agent-Based Computational Economics Simulation of Electricity Market CHEN Nai-shi, WANG Hai-ning, ZHOU Hai-ming, LI Wei-gang (China Electric Power Research Institute, Haidian District, Beijing 100192, China) ABSTRACT: ACE simulation of electricity market is an 力市场交易的实际情况，但易受干扰，得出的结果 important research method of electricity market, however, the 往往不收敛，不能对一些因素进行定性说明；而基 results of such a research on the agent simulation method are 于代理的仿真方法主要是用智能算法代理发电公 rarely reported. In this paper, the cooperative particle swarm 司进行竞价策略的选择，然后根据每次出清的结果 algorithm is applied to the intelligent agent based bidding, and 研究电力市场的各类问题。计算机仿真非常理性， by means of simulative test of electricity market, the agent process of cooperative particle swarm algorithm during the 尽管与实际市场情况有些差别，但能清晰地反映出 simulation is analyzed. Through the tests of various cost 每一个因素的影响，电力市场 ACE 仿真中的智能 variance and supply-demand level variance, the responses of 体是对人的学习与决策行为的模拟[1] 。目前对人的 cooperative particle swarm algorithm to cost factor and 学习能力的模拟主要采用遗传算法或强化学习的 supply-demand factor in the agent based bidding are analyzed 方法[2-4] 。文献[5]建立了基于基因算法的代理进化 and compared with the agent based results adopting Q-learning algorithm. From test results it can be seen that it i