中国电力 ›› 2025, Vol. 58 ›› Issue (3): 142-150.DOI: 10.11930/j.issn.1004-9649.202407038

• 新型电网 • 上一篇    下一篇

调控领域人工智能应用的高复用性验证平台

高志远1(), 庄卫金1(), 李峰1(), 于芳1(), 张鸿1(), 王艳1(), 夏旻2()   

  1. 1. 中国电力科学研究院有限公司(南京),江苏 南京 210003
    2. 江苏省大数据分析技术重点实验室(南京信息工程大学), 江苏 南京 210044
  • 收稿日期:2024-07-05 出版日期:2025-03-28 发布日期:2025-03-26
  • 作者简介:
    高志远(1972),男,通信作者,高级工程师(教授级),从事智能电网、厂站自动化系统、电力市场等研究,E-mail:gaozhiyuan@epri.sgcc.com.cn
    庄卫金( 1978) ,男,高级工程师,从事电力系统自动化研究,E-mail:zhuangweijin@epri.sgcc.com.cn
    李峰( 1976) ,男,高级工程师(研究员级),从事电力系统自动化、人工智能等研究,E-mail:lifeng2@epri.sgcc.com.cn
    于芳 (1985),女,硕士,高级工程师,从事电网调度自动化研究,E-mail:yufang@epri.sgcc.com.cn
    张鸿 (1980),男,硕士,高级工程师,从事电力系统自动化研究,E-mail:zhanghong@epri.sgcc.com.cn
    王艳 (1977),女,硕士,高级工程师,从事电网调度自动化研究,E-mail:wangyan@epri.sgcc.com.cn
    夏旻(1983),男,博士,教授,从事机器学习、大数据分析等研究,E-mail:xiamin@nuist.edu.cn
  • 基金资助:
    国家电网有限公司科技项目(5100-202255384A-2-0-ZN)。

High Reusability Verification Platform for Artificial Intelligence Applications in Power Grid Dispatching and Control Field

Zhiyuan GAO1(), Weijin ZHUANG1(), Feng LI1(), Fang YU1(), Hong ZHANG1(), Yan WANG1(), Min XIA2()   

  1. 1. China Electric Power Research Institute, Nanjing 210003, China
    2. Jiangsu Key Laboratory of Big Data Analysis Technology (Nanjing University of Information Science and Technology), Nanjing 210044, China
  • Received:2024-07-05 Online:2025-03-28 Published:2025-03-26
  • Supported by:
    This work is supported by Science and Technology Project of SGCC (No.5100-202255384A-2-0-ZN).

摘要:

近年来,电网调控领域智能化发展趋势明显,但由于人工智能(artificial intelligence,AI)技术本身的黑盒性质,正确性和可靠性验证问题日渐重要,能够支持各种不同人工智能模型验证的复用性支持平台尤为重要。通过对验证需求和问题的分析,提出一种包含基础设施、平台服务和应用服务的3层验证平台架构,分析了该架构下AI模型验证的运行模式,开展了案例和样本复用、容器化和镜像、多框架和算法集成等关键技术研究,并进行了开发和实施。该平台实现了算力和数据资源的统一管理和复用、兼容支持多种不同AI框架和算法库、验证场景的动态生成等,为后续调控领域人工智能技术推广应用奠定了基础。

关键词: 电网调控, 人工智能, 机器学习, 新型电力系统, 信息通信技术, 能量管理系统

Abstract:

In recent years, the trend of intelligent development in the field of power grid dispatching and control has been obvious. However, due to the black box nature of Artificial Intelligence (AI) technology itself, correctness and reliability verification is becoming increasingly important. Constructing a reusable supporting platform that can support verification of various AI models is particularly urgent. A three-layer verification platform architecture including infrastructure, platform services, and application services is thus proposed based on an analysis of the needs and issues. The operation mode of AI model validation under this architecture is analyzed, and the key technologies such as case and sample reusing, containerization and mirroring, as well as multi framework and algorithm integration are studied. Finally, the related development and implementation are also carried out. This platform has achieved unified management and reuse of computing power and data resources, compatibility with different AI frameworks and algorithm libraries, dynamic generation of verification scenarios, etc., laying a foundation for the future promotion and application of AI technology in the field of power grid dispatching.

Key words: power grid dispatching and control, artificial intelligence (AI), machine learning, new power system, information and communication technology (ICT), energy management system (EMS)