发电技术 ›› 2024, Vol. 45 ›› Issue (2): 353-362.DOI: 10.12096/j.2096-4528.pgt.22152

• 智能电网 • 上一篇    下一篇

基于长短期记忆神经网络的检修态电网低频振荡风险预测方法

付红军1, 朱劭璇2, 王步华1, 谢岩2, 熊浩清1, 唐晓骏2, 杜晓勇1, 李程昊3, 李晓萌3   

  1. 1.国网河南省电力公司,河南省 郑州市 450052
    2.中国电力科学研究院有限公司,北京市 海淀区 100192
    3.国网河南省电力公司电力科学研究院,河南省 郑州市 450052
  • 收稿日期:2023-05-05 出版日期:2024-04-30 发布日期:2024-04-29
  • 通讯作者: 朱劭璇
  • 作者简介:付红军(1968),男,教授级高工,研究方向为电力系统运行与控制,新能源发电及并网,电力储能应用;
    朱劭璇(1989),男,博士,工程师,研究方向为电力系统稳定分析与控制,本文通信作者,1024792840@qq.com
  • 基金资助:
    国网河南省电力公司科技项目(5217022000A8)

Risk Prediction Method of Low Frequency Oscillation in Maintenance Power Network Based on Long Short Term Memory Neural Network

Hongjun FU1, Shaoxuan ZHU2, Buhua WANG1, Yan XIE2, Haoqing XIONG1, Xiaojun TANG2, Xiaoyong DU1, Chenghao LI3, Xiaomeng LI3   

  1. 1.State Grid Henan Electric Power Company, Zhengzhou 450052, Henan Province, China
    2.China Electric Power Research Institute, Haidian District, Beijing 100192, China
    3.Electric Power Research Institute of State Grid Henan Electric Power Company, Zhengzhou 450052, Henan Province, China
  • Received:2023-05-05 Published:2024-04-30 Online:2024-04-29
  • Contact: Shaoxuan ZHU
  • Supported by:
    Science and Technology Foundation of State Grid Henan Electric Power Company(5217022000A8)

摘要:

随着电网规模扩大和电力元件不断增加,电力系统检修方式变得日趋复杂,仅依靠传统方法难以对海量检修方式下电网的低频振荡风险进行评估。针对此问题,提出了一种基于长短期记忆(long short term memory,LSTM)神经网络的检修态电网低频振荡风险预测方法。首先,提出了电力系统检修方式的统一编码方法,使计算机能够快速、准确识别电网在各种检修方式下的运行状态;然后,基于同步相量测量单元(phasor measurement unit,PMU)实时测量的电网历史运行数据,利用LSTM神经网络对不同检修方式下电网的低频振荡次数进行预测,从而评估检修态电网发生低频振荡的风险;最后,以华中地区某省级电网为算例,验证了所提方法的准确性和快速性。

关键词: 电力系统, 检修方式, 计算机编码, 低频振荡, 风险预测, 长短期记忆(LSTM)

Abstract:

With the expansion of power grid scale and the increase of power components, the maintenance methods of power system become more and more complex. It is difficult to evaluate the low-frequency oscillation risk of power grid under massive maintenance only by traditional methods. To solve this problem, a risk prediction method of low-frequency oscillation in maintenance power network based on long short term memory (LSTM) neural network was proposed. Firstly, the unified coding method of power system maintenance mode was proposed, so that the computer can quickly and accurately identify the operation state of power grid under various maintenance modes. Then, based on the historical data measured in real time by phasor measurement unit (PMU), the number of low-frequency oscillation of power grid under different maintenance modes was predicted by using LSTM neural network, so as to evaluate the risk of low-frequency oscillation of power grid under maintenance. Finally, a regional power grid in central China was taken as an example to verify the accuracy and rapidity of the proposed method.

Key words: power system, maintenance method, computer coding, low frequency oscillation, risk prediction, long short term memory (LSTM)

中图分类号: