基于一维卷积网络的工业设备故障智能诊断系统设计  

Design of industrial equipment fault intelligent diagnosis system based on one⁃dimensional convolutional networks

作　　者：李浩玮 王伟 王明朗 韦江平 黄智榜 LI Haowei;WANG Wei;WANG Minglang;WEI Jiangping;HUANG Zhibang(Longtan Hydroelectric Power Plant,Longtan Hydropower Development Co.,Ltd.,Nanning 530000,China)

机构地区：[1]龙滩水电开发有限公司龙滩水力发电厂,广西南宁530000

出　　处：《电子设计工程》2024年第22期96-100,共5页Electronic Design Engineering

基　　金：2022年龙滩水电开发有限公司龙滩水力发电厂信息化项目《基于桂冠电力生态云平台的智慧设备系统建设项目》(CDT-LTHPC-E-2395)。

摘　　要：为了对工业设备进行在线、离线软故障诊断,提出了一种基于数据驱动的方法,该方法是基于周期性突发注入和非侵入式电容耦合的一维卷积网络故障诊断新方法。使用非侵入式电容耦合提取软故障特征,利用不同脉冲序列作为周期入射信号,使用一维卷积网络模型学习高级特征,根据合成信号来消除盲点问题。实验结果表明,提出的故障智能诊断系统可以在在线和离线状态下进行故障诊断,模型搭建简单,易于实现,并且平均分类准确率均较高,适用于在实际生产,符合工业自动化发展需求。In order to diagnose the soft faults of industrial equipment online and offline,a data⁃driven method is proposed.This method is a new fault diagnosis method of one⁃dimensional convolutional networks based on periodic burst injection and non⁃invasive capacitance coupling.The soft fault features are extracted by non⁃invasive capacitive coupling,and different pulse sequences are used as periodic incident signals.The advanced features are learned by one⁃dimensional convolutional network model,and the blind spot problem is eliminated according to the synthesized signals.The experimental results show that the proposed fault intelligent diagnosis system can diagnose faults both online and offline.The model is simple,easy to implement,and the average classification accuracy is high.It is suitable for actual production and meets the development needs of industrial automation.

关 键 词：工业设备 故障智能诊断 一维卷积网络 电容耦合 电缆故障 

分 类 号：TN710[电子电信—电路与系统]