基于优势执行者-评论家算法的智能操作票生成方法  

Intelligent Operation Order Generation Method Based on the Advantage Actor Critic

作　　者：付乐勇 徐启峰[1] 黄奕钒 FU Le-yong;XU Qi-feng;HUANG Yi-fan(College of Electric Engineering and Automation,Fuzhou University,Fuzhou 350116,China)

机构地区：[1]福州大学电气工程与自动化学院,福建福州350116

出　　处：《电气开关》2023年第5期18-23,共6页Electric Switchgear

基　　金：国家自然科学基金项目(51977038);国家外国专家局“高端外国专家引进计划”(G2021020009L)。

摘　　要：目前人工或计算机辅助开票,均存在智能辅助手段不足的问题,无法满足电网智能化运行管理的要求。因此提出了一种基于优势行动者-评论家(A2C)算法的智能操作票生成方法,包括建立了以间隔为基本单元的变电站拓扑模型,然后基于SCADA系统的遥测、遥信等信息搭建智能体实时开票环境,以确保开票环境的真实性和实时性,最后设计了A2C算法的环境状态、动作选择与奖惩函数,并进行了实例验证。结果表明,所提方法能够可靠实现指定任务的操作票智能生成的目的,表明了该模型和算法的实用性。Now manual or computer-assisted billing has the problem of insufficient intelligent auxiliary means,which can′t meet the demands of power grid intelligent operation management.Therefore,this paper proposed an intelligent operation order generation method based on the Advantage Actor Critic(A2C)algorithm,which includes establishing a substation topology model with interval as the basic unit.And then building a real-time billing environment for Agent,based on the SCADA system′s telemetry,remote signaling and other information.Thus,it can ensure the authenticity and real-time nature of the billing environment for the Agent.Finally,the environment state,action selection and reward-penalty function of the A2C algorithm are designed,and examples can verify this method.The results show that the proposed method can reliably achieve the intelligent generation of operation order for specified tasks,which shows the practicability of the model and algorithm.

关 键 词：优势行动者-评论家 倒闸操作 操作票 强化学习 变电站 

分 类 号：TM734[电气工程—电力系统及自动化]