一种面向继保设备二次回路的接地故障LSTM精准定位技术研究
Research on LSTM-Based Precise Ground Fault Localization Technology for Secondary Circuits of Relay Protection Equipment
-
摘要: 针对变电站继保设备二次回路单相接地故障检测中的准确性与可靠性不足等问题,本文提出一种面向继保设备二次回路的接地故障LSTM精准定位技术。该方法利用改进的D-S证据理论,对源自电流、电压、暂态波形以及设备状态等多源信号进行有效融合,解决高冲突证据融合失真问题,构建起统一的决策框架;然后运用变分模态分解(VMD)技术对零序电流信号展开自适应分解,提取前4个本征模态函数(IMF)的瞬时参数作为关键特征,并通过Z-score标准化处理增强特征的可比性;最后,构建带辅助门控单元的改进 LSTM 分类器,发挥其门控机制的优势,强化微弱故障时序特征提取能力,实现对时序特征的精准建模与故障准确判别。应用结果表明,所提出的方法在测试集上展现出卓越的性能,故障识别率高达95%以上,漏检率低于1%,应用效果较好。Abstract: In response to the issues of insufficient accuracy and reliability in the detection of single-phase grounding faults in the secondary circuit of substation relay protection equipment, this paper proposes an LSTM accurate positioning technology for grounding faults in the secondary circuit of relay protection equipment. This method utilizes an improved D-S evidence theory to effectively fuse multi-source signals such as current, voltage, transient waveforms, and equipment status, solving the problem of high conflict evidence fusion distortion and constructing a unified decision-making framework; Then, the Variational Mode Decomposition (VMD) technique is applied to adaptively decompose the zero sequence current signal, extracting the instantaneous parameters of the first four Intrinsic Mode Functions (IMF) as key features, and enhancing the comparability of the features through Z-score standardization processing; Finally, an improved LSTM classifier with auxiliary gating units is constructed to leverage the advantages of its gating mechanism, enhance the ability to extract weak fault temporal features, and achieve accurate modeling of temporal features and accurate fault discrimination. The application results show that the proposed method exhibits excellent performance on the test set, with a fault recognition rate of over 95% and a missed detection rate of less than 1%. The application effect is good.
下载: