摘要
方式计算中衍生方式间的知识经验多数是可以复用的。该文基于一种经验轨迹空间驱动分析方法,结合局部电压调整经验的复用性,研究设计了可复用的电压调整场景及分析经验,提出了经验轨迹空间驱动的电压调整强化学习模型并改进Q-learning算法,实现了从分析经验轨迹中提取局部电压调整所需的调节资源空间以及学习如何组织数据实现电压控制目标。该方法基于对分析经验的记录和处理,解决了调整资源空间维度过大以及调整方案的可靠性问题,并且具备探索更优方案的能力,为其他潜在场景提供了参考,有望在传统人工智能技术的基础上进一步推动仿真分析的智能化。
Most of the knowledge and experience derived from the similar modes in the mode computing can be reused.Based on an experience-track-space driven analysis and the reusability of the partial voltage adjustment experience,this paper designs a reusable voltage adjustment scenario and analysis experience.It proposes a reinforcement learning model of the voltage adjustment driven by the experience-track space and improves the Q-learning algorithm to extract the adjustment resource space required for the partial voltage adjustment from the analysis experience trajectory and learn how to organize data to achieve the voltage control objectives.By recording and processing the analysis experience,this method solves the problem of the too large dimension of the adjustment resource space and the reliability of the adjustment scheme.Meanwhile this method has the ability to explore a better scheme,which provides a reference for the other potential scenarios.This method is expected to further promote the intellectualization of the simulation analysis on the basis of the traditional AI technology.
作者
于凯文
卜广全
吕晨
YU Kaiwen;BU Guangquan;LÜChen(China Electric Power Research Institute,Haidian District,Beijing 100192,China)
出处
《电网技术》
EI
CSCD
北大核心
2023年第7期2869-2878,共10页
Power System Technology
基金
国家重点研发计划资助项目“人在回路的大电网调控混合增强智能基础理论”(2018AAA0101500)。
关键词
经验轨迹空间
潮流计算
电压调整
强化学习
experience-track space
power flow calculation
voltage adjustment
reinforcement learning
