中国电力 ›› 2025, Vol. 58 ›› Issue (9): 68-78.DOI: 10.11930/j.issn.1004-9649.202502061

• 提升新能源和新型并网主体涉网安全能力关键技术 • 上一篇    下一篇

基于网络流算法和深度神经网络的智能变电站二次系统故障定位方法

陶军1(), 钟鸣1(), 张艺2(), 刘锋1, 武玉珠1(), 夏振兴3()   

  1. 1. 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司,内蒙古 呼和浩特 010020
    2. 内蒙古电力(集团)有限责任公司乌海超高压供电分公司,内蒙古 乌海 016000
    3. 武汉凯默电气有限公司,湖北 武汉 430223
  • 收稿日期:2025-02-26 发布日期:2025-09-26 出版日期:2025-09-28
  • 作者简介:
    陶军(1968),男,高级工程师(教授级),从事电力系统自动化研究,E-mail:taojun1968@163.com
    钟鸣 (1987),男,硕士,高级工程师,从事电力系统继电保护研究,E-mail:zhongming0005@163.com
    夏振兴(1995),男,通信作者,工程师,从事电力二次系统智能运维技术研究,E-mail:xiazhenxing@kemov.com
  • 基金资助:
    国家重点研发计划重点专项资助项目(2023YFB2405900);内蒙古电力(集团)有限责任公司2024年度科技项目(2024-4-50)。

Fault Location Method for Secondary System of Smart Substations Based on Network Flow Algorithm and Deep Neural Network

TAO Jun1(), ZHONG Ming1(), ZHANG Yi2(), LIU Feng1, WU Yuzhu1(), XIA Zhenxing3()   

  1. 1. Inner Mongolia Electric Power (Group) Co., Ltd., Inner Mongolia Electric Power Science Research Institute Branch, Hohhot 010020, China
    2. Inner Mongolia Electric Power (Group) Co., Ltd., Wuhai Ultra-high Voltage Power Supply Branch, Wuhai 016000, China
    3. Wuhan Kaimo Electric Co., Ltd., Wuhan 430223, China
  • Received:2025-02-26 Online:2025-09-26 Published:2025-09-28
  • Supported by:
    This work is supported by National Key Research and Development Program of China (No.2023YFB2405900), 2024 Science and Technology Project of Inner Mongolia Electric Power (Group) Co., Ltd. (No.2024-4-50).

摘要:

现有智能变电站二次系统故障定位方法依赖于特定类型的故障特征量,缺乏对多种故障类型综合处理的能力,在面临电力网络动态变化时,无法快速对方案进行修正。针对该挑战,提出了一种基于网络流算法和深度神经网络(deep neural network,DNN)的故障定位方法。采用新的故障类型分类方法,重新定义简单故障、伪复杂故障和复杂故障。构建故障特征编码与矩阵关系模型,并引入网络流算法,解决复杂故障定位中的链路故障与节点故障定位模糊的问题。将网络流算法与深度神经网络模型深度融合,实现对智能变电站二次系统故障的精准定位。通过仿真算例比较发现,所提方法不仅能够提高复杂故障识别的准确性,缩短故障定位时间,而且可以有效应对电力系统动态变化,提升了故障定位能力。

关键词: 智能变电站, 网络流算法, 深度神经网络, 故障定位

Abstract:

The existing intelligent substation secondary system fault location method relies on specific types of fault feature quantities, lacking the ability toively process multiple fault types. It is unable to quickly modify the scheme when facing dynamic changes in the power network. To address this challenge, a fault location method based on the network algorithm and deep neural network (DNN) is proposed. A new fault type classification method is adopted, and the simple fault, pseudo-complex fault, and complex fault are re. A fault feature coding and matrix relationship model are constructed, and the network flow algorithm is introduced to solve the problem of fuzzy positioning of link fault and node fault in complex fault. The network flow algorithm is deeply integrated with the deep neural network model to achieve accurate positioning of intelligent substation secondary system faults. Through simulation example comparison, it is found that the method can not only improve the accuracy of complex fault recognition and shorten the fault location time, but also effectively cope with the dynamic changes of the power system, and improve the fault location.

Key words: smart substation, network flow algorithm, deep neural network, fault location


AI


AI小编
您好!我是《中国电力》AI小编,有什么可以帮您的吗?