论文部分内容阅读
基于代理的计算经济学已经成为电力市场研究的一种重要方法,但是,作为一种新的方法,还有许多问题没有解决,其中一个基本的问题是研究方法的问题,包括仿真实验的设计方法、仿真结果的分析方法等。文中首先给出了电力市场仿真中一些关键概念的定义,然后以电力市场研究中常用的一种代理学习算法——Roth-Erev法为例,基于一个拥有4个发电商的日前电力市场,进行了市场模型、智能代理模型、实验目的及实验方案的设计。设计了3组实验,根据前2个实验的结果,对基于代理的电力市场仿真实验轮数的设定、实验盘数的设定等问题进行了分析和讨论,最后一个实验对Roth-Erev算法中一些参数的选取问题进行了探讨,展示了其在选取时的矛盾性,并从加强学习法的机理上分析了其原因。
Agent-based computational economics has become an important method in power market research. However, as a new method, there are still many problems that have not been solved. One of the basic problems is the problem of research methods, including the design of simulation experiments , Analysis of simulation results and so on. Firstly, the definition of some key concepts in power market simulation is given. Then, based on the Roth-Erev method, a proxy learning algorithm commonly used in power market research, based on a day-ahead electricity market with 4 power generators, The market model, intelligent agent model, the purpose of experiment and experimental design. According to the results of the first two experiments, this paper analyzes and discusses the setting of experiment-based power market simulation rounds and the setting of the number of experimental disks. The last experiment analyzes the Roth-Erev algorithm In the selection of some parameters were discussed, showing its contradiction in the selection, and to strengthen the learning mechanism from the analysis of its causes.