机器学习在智能配用电领域中的应用:北美工程实践概述

Application of Machine Learning in Field of Smart Power Distribution and Utilization:Overview of Engineering Practice in North America

机器学习技术是助力能源转型、促进清洁能源消纳的重要工具。近年来,机器学习技术在电力系统中的应用已得到广泛关注。由于机器学习技术的"黑箱"特征,使其在可解释性、鲁棒性等方面仍有待提升,与电力系统高可靠性的运行要求存在一定矛盾,导致其实际工程应用滞后于理论研究。对于机器学习技术的实际应用情况,文中聚焦于北美地区配用电领域,从源、网、荷3个角度梳理了机器学习技术的典型工程实践项目,概述了每个项目的方法、效果以及从中得到的启示。进一步地,将以上项目归纳为态势感知、决策支持2个类别共计5个应用场景,并从工程实践角度分析了下阶段机器学习技术的研究方向。

Machine learning technique is important for assisting the energy transition and promoting the renewable energy consumption.In recent years,the application of machine learning technique in power systems has been widely concerned.Due to the‘black box’nature of machine learning technique,its interpretability and robustness are still to be improved.And there is a certain contradiction with the operation requirements of high reliability in the power system,which leads to its practical engineering application lagging behind the theoretical research.In order to introduce the practical application of machine learning technique,this paper focuses on the field of power distribution in North America.The typical engineering practice projects of machine learning technique are summarized from the perspectives of source,network and load,and the method,effect and inspiration of each project are also outlined.Further,the above projects are classified into two categories,i.e.,situational awareness and decision support,and a total of five application scenarios.And the research areas of machine learning technique in the next stage is analyzed from the perspective of engineering practice.