摘要
变压器作为供电系统的枢纽设备,其状态直接影响电力供应的稳定,因此对变压器进行状态监测和故障诊断具有重要意义。针对传统的监测方法如油色谱分析无法对变压器机械故障实现监测,提出了一种基于深度学习的变压器声纹故障诊断方法。通过采集变压器的声纹信号,采用短时傅里叶变换(STFT)提取声纹的时频特征,将其转换为二维特征图谱并输入卷积神经网络(CNN)进行分类训练与验证。该模型在直流偏磁、风机老化及正常运行的三类工况中,均表现出良好的分类准确性。
Transformer as the core equipment of the power supply system,the state of its equipment directly affects the stability of power supply.Therefore,condition monitoring and fault diagnosis of transformers are of great significance.However,traditional monitoring methods such as oil chromatography analysis cannot monitor the mechanical faults of transformers.To address this problem,this paper proposes a deep learning-based transformer acoustic fault diagnosis method.The acoustic pattern signal of the transformer is collected and the time-frequency features of the acoustic pattern are extracted by using the short-time Fourier transform(STFT),which is converted into a two-dimensional feature map and inputted into a convolutional neural network(CNN)for classification training and validation.The model designed in this paper shows good classification accuracy in three types of working conditions under different operating conditions,including DC bias,fan aging and normal operation.
作者
李波
闫胜春
LI Bo;YAN Shengchun(Guoneng Shuohuang Railway Development Co.,Ltd.,Suning 062350,China)
出处
《电工技术》
2025年第17期60-62,65,共4页
Electric Engineering
基金
国能朔黄铁路发展责任有限公司科技创新项目(编号SHTL-24-42)。
关键词
深度学习
变压器
声纹识别
故障诊断
deep learning
transformer
voiceprint recognition
fault diagnosis
