基于1D-CNN的振弦传感器故障诊断及修复系统  

Fault diagnosis and restoration system for vibrating wire sensor based on 1D-CNN

作　　者：王祥安 王辉[2] 李永红 李钊 景波云 刘艳平 WANG Xiang'an;WANG Hui;LI Yonghong;LI Zhao;JING Boyun;LIU Yanping(NARI Technology Co.,Ltd.,Jiangsu Nanjing 211106,China;China Yangtze Power Co.,Ltd.,Hubei Yichang 443002,China)

机构地区：[1]国电南瑞科技股份有限公司,江苏南京211106 [2]中国长江电力股份有限公司,湖北宜昌443002

出　　处：《工业仪表与自动化装置》2025年第2期87-91,共5页Industrial Instrumentation & Automation

基　　金：该成果由长江电力股份有限公司项目Z242302017资助。

摘　　要：针对振弦传感器在应力监测过程中,受到埋设不良、接线过长、激振不足等影响,可能会无法准确测量的问题,提出了一种基于一维卷积神经网络(1D-CNN)的振弦传感器故障诊断方法,以振弦传感器输出信号幅值为输入,能快速准确诊断故障。同时,采用短时傅里叶变换,找到信号中的衰减分量,实现了对一种振弦传感器故障的修复,使得传感器重新投入运行。最后构建了振弦传感器的故障预测与健康管理(Prognostics and Health Management,PHM)系统,对振弦传感器故障识别、诊断与修复具有一定意义。In the stress monitoring process,the vibrating wire sensor may fail to measure accurately due to poor installation,excessive wiring length,and inadequate excitation.A vibrating wire sensor fault diagnosis method based on a one-dimensional convolutional neural network(1D-CNN)has been introduced to assess the sensor's functionality and pinpoint instrument malfunctions.The method utilizes the output signal amplitude of the vibrating wire sensor as input,enabling rapid and precise fault diagnosis.Additionally,employing the short-time Fourier transform,the decay components within the signal were identified,facilitating the repair of a vibrating wire sensor fault and restoring the sensor to operation.Ultimately,a Prognostics and Health Management(PHM)system for vibrating wire sensors has been developed,significantly aiding in the identification,diagnosis,and repair of sensor faults.

关 键 词：振弦传感器 一维卷积神经网络 短时傅里叶变换 故障预测与健康管理系统 

分 类 号：TH823[机械工程—仪器科学与技术]