滑动窗口协议
故障检测与隔离
波形
计算机科学
光学(聚焦)
可靠性(半导体)
断层(地质)
控制理论(社会学)
实时计算
事件(粒子物理)
电阻抗
高阻抗
电子工程
探测理论
阶跃检测
工程类
窗口(计算)
算法
故障指示器
探测器
信号(编程语言)
模式识别(心理学)
钥匙(锁)
变更检测
分布(数学)
人工智能
基础(线性代数)
时频分析
灵敏度(控制系统)
作者
Chu Tong,Mou‐Fa Guo,Jian-Hong Gao,Cui Hong
出处
期刊:
[Institute of Electrical and Electronics Engineers]
日期:2025-11-26
卷期号:: 293-297
标识
DOI:10.1109/psre67836.2025.11397936
摘要
Existing high-impedance fault (HIF) detection methods suffer from low detection accuracy due to the insufficient utilization of the non-periodicity and long duration characteristics of HIF waveforms. In this paper, we propose a TimesNet-based HIF detection method in distribution networks. TimesNet decouples the time sequence, allowing for a focus on both periodic and non-periodic waveforms, thereby addressing the issue of underutilized non-periodic features in existing methods. Once a fault was triggered by triggering algorithm, the zero-sequence current in a certain length would be captured and traversed by a sliding window. With the movement of the sliding window, the overlapped waveform would be decoupled by TimesNet into different cycles, with attention paid to the temporal variations both weekly and within cycles to achieve the HIF detection at the level of sliding window. When the obtained zero-sequence current is processed using a sliding window, the corresponding detection results at the sliding window level serve as the basis for the HIF detection results at the event level. The proportion of detection results at the sliding window level is calculated and compared with a threshold to determine whether the event as a whole constitutes a HIF. Simulation and experimental validation demonstrate that the proposed meth-od exhibits effectiveness and reliability in detecting HIF.
科研通智能强力驱动
Strongly Powered by AbleSci AI