基于深度学习的电力系统低频振荡模式评估  

Deep learning-based low-frequency oscillation mode assessment for power systems

作　　者：朱美晔 ZHU Meiye(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》2024年第12期24-31,共8页Electrotechnical Application

基　　金：吉林省国际科技合作项目(20230402074GH)。

摘　　要：随着新型电力系统发展,小干扰稳定问题日益严重,准确评估电力系统低频振荡模式对于提高电网稳定性具有重要意义。提出了一种基于卷积神经网络的电力系统低频振荡模式评估模型,与传统的基于稳态信息的评估方法不同,所提方法考虑了系统运行时负荷受到随机扰动的影响,利用广域测量系统监测变量和关键振荡模式阻尼比并作为模型的输入和输出,采用卷积神经网络训练得到其映射关系。IEEE 10机39节点算例的实验表明,该模型经过离线训练后,能够准确计算系统关键振荡模式阻尼比,并且具有较强的抗干扰能力。With the development of new power systems,the problem of small disturbance stabilization has become increasingly serious,and accurate assessment of low-frequency oscillation modes of power systems is of great significance for improving the stability of power grids.Different from the traditional low-frequency oscillation mode assessment method based on steady state information,the paper considers the influence of random disturbances on the load during system operation,utilizes the Wide Area Measurement System(WAMS)monitoring variables and the key oscillation mode damping ratio as the input and output of the model,and adopts convolutional neural network training to obtain its mapping relationship,and proposes a low-frequency oscillation mode assessment model for power systems based on convolutional neural network.Experiments on the IEEE10 machine 39-node algorithm show that the model can accurately calculate the system critical oscillation mode damping ratio after offline training and has strong anti-interference ability.

关 键 词：低频振荡评估 卷积神经网络 随机扰动 样本集 

分 类 号：TM712[电气工程—电力系统及自动化] TP18[自动化与计算机技术—控制理论与控制工程]