深度学习在电力系统频率分析与控制中的应用综述

Review on Deep Learning Applications in Power System Frequency Analysis and Control

低惯量可再生能源的发展及大电网复杂互联导致现代电力系统呈现新特征和新形态,具有典型的强时变性、强非线性、随机不确定性、数据多样性、局部可观测性等特征,电力系统分析方法面临更大挑战。深度学习作为一项新的机器学习技术路径,其强大的数据分析、预测、分类能力在电力系统频率分析与控制等复杂问题中具有独特优势。首先分析深度学习的基本原理与研究进展,介绍深度学习的训练方法、典型模型结构及应用特点,总结频率动态趋势感知、频率安全与稳定评估以及频率控制与调节领域的问题特征及深度学习的应用现状,并探讨针对每类问题深度学习应用的适应性,构建了深度学习在电力系统频率分析与控制中的应用框架。最后,对深度学习的发展趋势及其在电力系统频率问题中的应用前景进行展望。

The development of low-inertia renewable energy and the complex interconnection of large power grids have led to new features and forms of modern power systems exhibiting,which are typically characterized by time-varying nonlinearity,uncertainty,data diversity,and local observability,etc.The problems and methods of power system analysis are becoming more complex.As a new technology path of machine learning,the deep learning(DL)has unique advantages in solving complex problems such as power system frequency analysis and control due to its powerful ability of data analysis,prediction,and classification.First,this paper analyzed the basic principle and research progress of DL,the training methods,typical model structures,and application features of DL were introduced.Secondly,the application status of DL in frequency dynamic situation awareness,frequency stability,and safety assessment,frequency control and regulation were summarized,and the adaptability of DL application to each kind of problem was discussed.Then,the application framework of DL in power system frequency analysis and control was constructed.Finally,the development trend of DL and its application in power system frequency were prospected.