基于卷积神经网络的电力系统低频振荡类型判别  

A convolutional neural network-based method for classification of power system electromechanical oscillation

作　　者：于泓智 YU Hongzhi(School of Electrical Engineering,Northeast Electric Power University,Jilin 132012,China)

机构地区：[1]东北电力大学电气工程学院,吉林吉林132012

出　　处：《电气应用》2023年第7期18-24,共7页Electrotechnical Application

摘　　要：低频振荡严重威胁电力系统的安全与稳定。电力系统低频振荡分为强迫振荡和自由振荡两种类型,由于振荡产生机理不同,对应所采取的抑制手段也不相同。低频振荡类型的判别对于后续合理采取抑制措施、提高电网稳定性具有重要意义。与此同时,深度学习在电力系统应用方向上得到了深远的发展,提出一种基于卷积神经网络的电力系统低频振荡类型判定方法。应用于10机39节点系统,与一般的数据驱动方法不同,采用端口供给能量作为特征输入,具有物理可解释的特性,用卷积神经网络对振荡端口供给能量数据分类代替端口供给能量的分解判别过程,准确快速地实现低频振荡类型的判别。Low frequency oscillation seriously threatens the security and stability of the power system.Due to the different mechanisms,the low-frequency oscillation of the power system is divided into forced oscillation and free oscillation,and the corresponding means are different.It is of great significance to identify the type of oscillation to reasonably take suppression measures and improve the stability of the power grid.At the same time,deep learning has made far-reaching development in the direction of power system application.A low-frequency oscillation type determination method for power system based on convolutional neural network classifier.Unlike general data-driven methods,this classifier has the characteristics of physical interpretation;port supplied energy is used as feature input,the scheme is verified by simulating the low frequency oscillation type of IEEE 39-bus power systems.

关 键 词：卷积神经网络 低频振荡分类 端口供给能量 特征提取 

分 类 号：TM712[电气工程—电力系统及自动化] TP183[自动化与计算机技术—控制理论与控制工程]