Study of Discharge Sound Diagnosis Based on Machine Learning and Convolutional Neural Networks

SUN Hanwen; LI Zhe; SHENG Gehao; JIANG Xiuchen

doi:10.13296/j.1001-1609.hva.2020.09.016

您当前的位置：

首页 >

文章列表页 >

Study of Discharge Sound Diagnosis Based on Machine Learning and Convolutional Neural Networks

更新时间：2025-04-26

- Study of Discharge Sound Diagnosis Based on Machine Learning and Convolutional Neural Networks
- High VoltageApparatus Vol. 56, Issue 9, Pages: 107-113(2020)
- 作者机构：
  
  上海交通大学电气工程系
- 作者简介：
- 基金信息：
- DOI：10.13296/j.1001-1609.hva.2020.09.016
  CLC： TP18;TM855
- Published：2020
- 稿件说明：
移动端阅览
[1]孙汉文,李喆,盛戈皞,江秀臣.基于机器学习与卷积神经网络的放电声音识别研究[J].高压电器,2020,56(09):107-113.

SUN Hanwen, LI Zhe, SHENG Gehao, et al. Study of Discharge Sound Diagnosis Based on Machine Learning and Convolutional Neural Networks[J]. High voltageapparatus, 2020, 56(9): 107-113.
[1]孙汉文,李喆,盛戈皞,江秀臣.基于机器学习与卷积神经网络的放电声音识别研究[J].高压电器,2020,56(09):107-113. DOI： 10.13296/j.1001-1609.hva.2020.09.016.

SUN Hanwen, LI Zhe, SHENG Gehao, et al. Study of Discharge Sound Diagnosis Based on Machine Learning and Convolutional Neural Networks[J]. High voltageapparatus, 2020, 56(9): 107-113. DOI： 10.13296/j.1001-1609.hva.2020.09.016.

摘要

为了实现对电气设备放电声音的精准检测

文中筛选比较了多种经典的机器学习算法和新兴的卷积神经网络算法

以期得到识别效果最优的选择。首先对音频进行预处理

再通过将放电声与环境噪声和变电站正常工况背景声混合来模拟变电站真实工作环境

并使用梅尔频率倒谱系数提取特征

最后采用支持向量机等机器学习算法与卷积神经网络算法进行识别

选取识别效果最佳的算法并考察不同采样频率、采样时长等因素对识别效果的影响。实验结果表明

使用梅尔频率倒谱系数提取特征可以良好区分放电与环境噪声

支持向量机在一系列算法中识别放电声音能力最强

采样频率、标准化方式等因素对识别效果影响较小。

Abstract

In order to realize more accurate detection of the discharge sound of electrical equipment

a variety of classical machine learning algorithms and the emerging convolution neural network algorithm were compared and screened to get the best choice.In this study

audio preprocessing was implemented firstly

then the discharge sound was mixed with environment noise and the background sound of substation on normal working condition to simulate the real working environment of substation.The feature vector and spectrum were extracted by Mel frequency cepstrum coefficients

and the recognition performances of convolution neural network and machine learning algorithms such as support vector machine（SVM） were compared.Selected the best recognition algorithm

the influence of different sampling rate or other factors were also studied.The experimental results show that the method of extracting eigenvector or spectrogram by Mel frequency cepstrum coefficient can distinguish discharge from environment noise

and SVM has the best performance in a series of algorithms.Sampling rate

methods of standardization and other factors do not have a significant influence on recognition accuracy.

关键词

Keywords

references

Views

322

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Multi-signal Joint Identification and Intelligent Diagnosis of High Voltage Circuit Breaker Faults Based on VMD-BiLSTM

Design of a Digital Twin-oriented Integrated Monitoring System for Substation Switchgear

Fault Diagnosis of High Voltage Disconnector Based on Attitude Angle and Speed Curves

Fault Diagnosis Method of Generator Based on Stacked Denoising Autoencoder Network and Multi-source Data Weighted Fusion

Intelligent Checking Method of Abnormal Electricity Data Based on Stacking Strategy and Deep Learning

Related Author

ZHANG Jinbo

ZHANG Zhaowei

WU Jianwen

SHAO Yang

LIANG Chuantao

SHEN Xiaosong

HU Dazheng

ZHANG Yiming

Related Institution

Xi’an High Voltage Apparatus Research Institute Co., Ltd.

School of Automation Science and Electrical Engineering, Beihang University

Shandong Taikai High Voltage Switchgear Co., Ltd.

Pinggao Group Co., Ltd.

Henan Pinggao Electric Co., Ltd.

⁰