东篱科研大数据发现系统（DRDS）

位置：成果数据库 > 期刊 > 期刊详情页

基于启发式动态规划方法的发电商最优竞价策略

期刊名称：电网技术．33（11）．71-75，2009．（中文核心）（第一基金）
时间：0
分类：F416.2[经济管理—产业经济]
作者机构：[1]四川大学水电学院,四川省成都市610065
相关基金：国家自然科学基金资助项目（50539140）.
相关项目：市场条件下流域梯级水电能源联合优化运行和管理的先进理论与方法

关键词：电力市场, 启发式动态规划, 近似动态规划, 智能体, electricity market, heuristic dynamic programming, approximate dynamic programming, agent

中文摘要：

电力市场环境下的购销竞价策略是一个复杂的动态问题，传统的数学解析方法很难对其进行分析计算，这在中长期交易时间段尤为突出。文章提出了以最优潮流为基础的双层中长期最优竞价策略学习模型。采用启发式动态规划方法，将动态规划的策略行为和奖惩因子迭代过程看作智能体（agent），将外界不确定因素当作该智能体的“外部环境”。在所处的环境条件下，agent通过评价环境作出判断来选择可行的策略方案，通过学习以往的报价经验和对手的行为来指导自身达到最优生产效益。在标准IEEE5节点6支路系统中进行实例计算，证明了所提出方法的适应性和优越性。

英文摘要：

The bidding strategy for electricity purchase and sale in electricity market environment is a complex dynamic problem that is especially prominent in the duration of medium- and long-term transaction, so it is hard for traditional mathematical analysis method to analyze and calculate such a problem. An optimal power flow based learning model for two-layer medium- and long-term optimal bidding strategy is proposed. Using heuristic dynamic programming method, the strategic behavior of dynamic programming and iteration of reward and penalty factors are regarded as agent and the external uncertain factors as the external environment of the agent. Under this condition, through the assessment and judgment on environment, the agent chooses feasible strategic scheme guides itself to attain optimal production revenue by learning former bidding experiences and the behavior of competitors. Taking IEEE 5-bus 6-branch test system as the case, calculation results prove that the proposed method is adaptive and superior.