人机混合增强决策智能在新型电力系统调控中的应用与展望

扫码查看

原文链接

国家科技期刊平台
NETL
NSTL
万方数据

中文摘要：新型电力系统的调控运行面临规模及随机性升高、海量多元资源协同困难等难题，现有传统优化和人工经验的调控手段难以应对。以强化学习为代表的决策智能技术在表征能力和决策速度方面具备显著优势，但在应用中存在诸多关键瓶颈。而人机混合增强智能(hybrid human-machine intelligence，HHMI)技术具有突破这些瓶颈、支撑高效智能调控的巨大潜力。目前HHMI理论研究尚处于初期，应用不足。为挖掘HHMI的潜力并探索其应用方案，基于新型电力系统调控与应急调整的实际需求，分析决策智能的优势与局限，在此基础上，论述HHMI的总体框架、关键技术、应用方案以及在实际应用中面临的关键问题，并对其未来的实践进行展望，可为HHMI在新型电力系统调控中的研究与应用提供思路。

外文标题：Application and Prospect of Hybrid Human-machine Decision Intelligence in the Dispatch and Control of Next-era Power Systems

外文摘要：The dispatch and operation of next-era power system faces challenges such as increased system scale and randomness,and difficulty in coordinating massive and diverse resources.Existing optimization-based and manual dispatch methods are difficult to tackle these challenges.Although decision intelligence represented by reinforcement learning has significant advantages in representation capability and decision speed,its application in power systems still faces critical bottlenecks.The hybrid human-machine intelligence(HHMI)has great potential to break through these bottlenecks and support the efficient and intelligent system dispatch and adjustments.At present,the research on HHMI is still in its infancy and its application lacks.To explore the potential and applications of HHMI,considering the needs of power system dispatch and adjustments,the advantages and limitations of decision intelligence are analyzed.On this basis,the overall framework,key methods,application solutions,and key problems in the practical application of HHMI are discussed,and its future practices are prospected.This provides ideas for the research and application of HHMI in the dispatch and control of next-era power systems.

外文关键词：

hybrid human-machinehuman-in-the-loophuman-computer interactionpower system dispatchgrid operation & controlintelligent decisionreinforcement learning

作者：

李鹏、黄文琦、梁凌宇、戴珍、曹尚、车亮、涂春鸣

展开 >

作者单位：

南方电网数字电网研究院有限公司,广东省广州市 510000

湖南大学电气与信息工程学院,湖南省长沙市 410082

关键词：

人机混合人在回路人机交互电力系统调度电网运行控制智能决策强化学习

基金：

国家重点研发计划项目南方电网数字电网研究院科技项目

项目编号：

2022YFB2403500670000KK52220002

出版年：

2024

DOI：

10.13334/j.0258-8013.pcsee.231797

中国电机工程学报

中国电机工程学会

中国电机工程学报

CSTPCD北大核心

影响因子：2.712

ISSN：0258-8013

年,卷(期)：2024.44(16)