面向新型电力系统运行的数据-物理融合建模综述  

Review of Hybrid Data-driven and Physics-based Modeling for the Operation of New-type Power Systems

作　　者：阮广春 何一鎏 谭振飞 钟海旺[1] RUAN Guangchun;HE Yiliu;TAN Zhenfei;ZHONG Haiwang(Department of Electrical Engineering,Tsinghua University,Haidian District,Beijing 100084,China)

机构地区：[1]清华大学电机工程与应用电子技术系,北京市海淀区100084

出　　处：《中国电机工程学报》2024年第13期5021-5036,I0001,共17页Proceedings of the CSEE

基　　金：国家重点研发计划项目(2020YFB0905900)。

摘　　要：构建新型电力系统是我国实现碳达峰、碳中和目标的关键,将给电力工业带来深刻变革与挑战。数据-物理融合建模(简称融合建模)是一类新兴的建模技术,能够同时发挥物理机理与数据的价值,有望成为新型电力系统重要的分析工具。为此,该文首先梳理融合建模的相关概念与应用场景,讨论近年来国内外的研究趋势与热点。进而从技术特征和融合模式两方面,提出针对融合模型的分析框架。同时,聚焦于新型电力系统运行领域,全方位总结整理融合建模在应对现有技术挑战方面的潜在优势及不足,并展望未来研究与工程实践的重点发展领域。Constructing new-generation power systems dominated by renewable energy is crucial to achieve China’s carbon neutrality goal,but this will inevitably bring significant changes and challenges to the existing power grids.Hybrid data-driven and physics-based modeling(hybrid modeling for short)is an emerging technique to combine the advantages of physic laws and data,showing great potential to serve as an important analysis tool for the new-generation power systems.To this end,this paper clarifies the relevant concepts and use cases at first,and then discusses the research trends and hotspots in recent literature.A general framework to evaluate the performance of hybrid modeling from an aspect of technical features and hybrid patterns is also proposed.In addition,this paper is focused on the operation of newgeneration power systems,and has fully summarized the pros and cons of hybrid modeling in addressing the existing technical challenges.Further suggestions for future research work and pilot projects are discussed at last.

关 键 词：新型电力系统 高比例可再生能源 数据驱动 知识驱动 人工智能 机器学习 

分 类 号：TM73[电气工程—电力系统及自动化]