基于声纹识别的油浸式变压器局部放电故障诊断研究  

Research on Fault Diagnosis of Partial Discharge in Oil-Immersed Transformers Based on Voice Recognition

作　　者：王理丽 李子彬 李军 王生杰 李秋阳 王子乐 杨潇洁 WANG Lili;LI Zibin;LI Jun;WANG Shengjie;LI Qiuyang;WANG Zile;YANG Xiaojie

机构地区：[1]国网青海省电力公司电力科学研究院,青海西宁810008

出　　处：《青海电力》2024年第4期36-41,共6页Qinghai Electric Power

摘　　要：油浸式变压器的局部放电会导致局部绝缘的损坏,并随着时间的推移逐渐扩大,最终导致设备的损坏或故障。为有效监测油浸式变压器的局部放电故障,以声纹识别技术为监测手段,对油浸式变压器局部放电产生的声音信号进行采集和处理,搭建局部放电故障诊断模型,实现变压器内部尖端放电、悬浮放电等5种局部放电故障的监测识别,同时以实际声纹故障数据进行模型训练和测试,有效提高识别精度,具有较高的实用价值和应用前景,可为电力系统的安全运行提供有效的技术支持和保障。Partial discharge in oil-immersed transformers can lead to local insulation damage,which gradually expands over time,eventually causing equipment damage or.In order to effectively monitor partial discharge failures in oilimmersed transformers,a monitoring method based on voiceprint recognition technology is proposed.The sound signals generated partial discharges in oil-immersed transformers are collected and processed,and a partial discharge failure diagnosis model is established to achieve the monitoring and identification of five types of discharge failures,such as internal tip discharge and floating discharge.Meanwhile,the model is trained and tested with actual voiceprint failure data,which effectively improves the accuracy identification.This method has high practical value and application prospects,and can provide effective technical support and guarantee for the safe operation of power systems.

关 键 词：变压器 声纹故障诊断 注意力机制 残差神经网络 梅尔语谱图 

分 类 号：TM406[电气工程—电器]