摘要
文中提出一种基于声发射技术的燃气调压器故障诊断方法,使用小波包分解对调压器声发射信号进行特征提取,利用BP神经网络模型对调压器进行早期的故障诊断。实验结果证明了这种方法的可行性,对调压器故障诊断的准确率达到95%以上。除此之外,ms级的诊断周期也让这种方法能够在较短时间内对调压器的运行状态做出准确判断,在故障发生时迅速做出反应,从而避免事故的发生。
In this paper, a method of gas pressure regulator fault diagnosis based on acoustic emission was proposed. Thewavelet packet decomposition was used for feature extraction of the acoustic emission signal. Then, based on the model obtainedin the BP neural network, early regulator fault can be detected. The experimental results prove the feasibility of this method. Theaccuracy rate of fault diagnosis of the pressure regulator is above 95%. In addition, the millisecond diagnostic cycle also makesthis method make an accurate judgment of the operation state of the pressure regulator in a short time, to react quickly when afault occurs to avoid an accident.
作者
刘瑶
位亚鹏
邢琳琳
陈涛涛
万仲飞
马治国
LIU Yao;WEI Ya-peng;XING Lin-lin;CHEN Tao-tao;WAN Zhong-fei;MA Zhi-guo(Beijing Gas Group Co.,Ltd.,Beijing 100011,China;School of Optics and Photonics,Beijing Institute of Technology,Beijing 100081,China;Institute of Acoustics,Chinese Academy of Sciences,Beijing 100190,China)
出处
《管道技术与设备》
CAS
2018年第6期25-28,共4页
Pipeline Technique and Equipment
基金
北京市燃气集团高压管网分公司支持项目(15JK015)
关键词
调压器
声发射
小波包
特征提取
BP神经网络
gas pressure regulator
acoustic emission
wavelet packet
feature extraction
BP neural network
