首页|可解释机器学习在电网调控领域中的应用

可解释机器学习在电网调控领域中的应用

扫码查看
原文链接
  • 国家科技期刊平台
  • NETL
  • NSTL
  • 万方数据
数据驱动的机器学习是新一代人工智能的核心技术,尽管该技术已经在电网调控领域取得了显著成果,但是可解释性差,阻碍了其在对安全可靠性要求极高的电网调控领域的实际工程应用.因此,提升电网调控领域机器学习技术的可解释性对提高其实用性至关重要.首先,从电网调度运行人员的角度,分析了机器学习可解释性的定义、目标和意义;然后,提出考虑可解释性的机器学习在电网调控领域应用的流程,介绍了典型的机器学习解释技术及其在电力系统预测和稳定评估场景的应用,通过实际案例验证了该技术在电网调控领域应用的可行性;最后,对电网调控领域机器学习可解释技术面临的挑战进行了分析和展望.通过该研究,为解决电网调控领域机器学习应用的不可解释问题提供思路和参考,进一步促进机器学习技术在该领域的实际工程应用.
Application of Interpretable Machine Learning in Power Grid Dispatching and Control
Data driven machine learning is the core technology of the new generation of artificial intelligence.Although this technology has achieved significant results in the field of power grid dispatching and control,its poor interpretability hinders its practical engineering application in the field of power grid dispatching and control,which requires extremely high safety and reliability.Therefore,improving the interpretability of machine learning technology in the field of power grid dispatching and control is crucial for enhancing its practicality.Firstly,from the perspective of power grid dispatch operators,the definition,objectives,and significance of machine learning interpretability are analyzed.Then,the process of applying machine learning considering interpretability in the field of power grid dispatching and control is proposed,and typical machine learning interpretation techniques and their applications in power system prediction and stability assessment scenarios are introduced.The feasibility of applying this technology in the field of power grid dispatching and control is verified through practical cases.Finally,an analysis and outlook are conducted on the challenges faced by machine learning interpretable technologies in the field of power grid dispatching and control.Through this study,ideas and references are provided to solve the inexplicable problem of machine learning applications in the field of power grid dispatching and control,further promoting the practical engineering application of machine learning technology in this field.

artificial intelligencemachine learningpower grid dispatching and controlinterpretable machine learningdata driven methodload forecastingpower system stability assessment

庞传军、王珅、余建明

展开 >

南瑞集团有限公司(国网电力科学研究院有限公司),江苏省 南京市 211106

北京科东电力控制系统有限责任公司,北京市 海淀区 100192

国家电网公司国家电力调度控制中心,北京市 西城区 100031

人工智能 机器学习 电网调度运行 可解释机器学习 数据驱动方法 负荷预测 电力系统稳定评估

国家电网总部科技项目

5108-202218280A-2-277-XG

2024

10.16543/j.2095-641x.electric.power.ict.2024.05.01

电力信息与通信技术
中国电力科学研究院

电力信息与通信技术

CSTPCD
影响因子：0.699
ISSN：1672-4844
年,卷(期)：2024.22(5)
  • 55