基于改进VMD与Transformer的GIS局部放电信号识别  

GIS Partial Discharge Signal Recognition Based on Improved VMD and Transformer

作　　者：苑超 丁晓森 朱雷 胡天羽 申张亮 马宏忠[2] YUAN Chao;DING Xiaosen;ZHU Lei;HU Tianyu;SHEN Zhangliang;MA Hongzhong(Nanjing Power Supply Company,State Grid Jiangsu Electric Power Co.,Ltd.,Nanjing 210002,China;School of Electrical and Power Engineering,Hohai University,Nanjing 211100,China)

机构地区：[1]国网江苏省电力有限公司南京供电分公司,江苏南京210002 [2]河海大学电气与动力工程学院,江苏南京211100

出　　处：《机械与电子》2025年第4期27-34,39,共9页Machinery & Electronics

基　　金：江苏省电力有限公司重点科技项目(J2023027)。

摘　　要：针对GIS局部放电识别方法在特征选择上处理过程复杂且识别准确率相对较低的问题,提出了一种基于改进海马优化变分模态分解(ISHO-VMD)和改进Transformer的局部放电信号识别方法。首先,通过ISHO-VMD对含噪声的局部放电信号进行分解,提取噪声最少的信号作为模型输入,从而降低识别过程中的计算复杂度,并避免了传统特征提取方法中可能出现的信息缺失。然后,通过改进Transformer的自适应特征捕捉能力,实现对不同局部放电类型的高精度识别。实验结果表明,所提方法显著提升了GIS局部放电类型识别的准确性,为GIS故障诊断提供了新思路。A partial discharge signal recognition method based on improved seahorse optimization variational mode decomposition(ISHO-VMD)and improved Transformer is proposed to address the problem of complex feature selection and relatively low recognition accuracy in GIS partial discharge recognition methods.First,ISHO-VMD decomposes noisy PD signals,isolating the least noisy component as model input,thereby reducing computational complexity and mitigating information loss inherent in traditional feature extraction.Then,an enhanced Transformer improves adaptive feature capture,enabling high precision recognition of different PD types.Experimental results demonstrate that the proposed method significantly boosts GIS PD recognition accuracy,offering a novel approach to GIS fault diagnosis.

关 键 词：气体绝缘开关 局部放电 注意力机制 优化算法 TRANSFORMER 

分 类 号：TM93[电气工程—电力电子与电力传动]