基于格拉姆角场边缘实现的轴承故障诊断系统
Bearing Fault Diagnosis System Based on Edge Implementation of Gramian Angle Field
屈绍宇 1李沅 1卢研宏1
作者信息
摘要
为解决2D-CNN轴承故障诊断前,将一维信号编码为二维特征矩阵导致的模型部署流程复杂和采集端资源浪费的问题,设计了一种基于格拉姆角场边缘实现的轴承故障诊断系统.系统以ZYNQ为核心,通过PL(可编程逻辑)端完成了格拉姆角场编码等信号预处理,采用以太网将特征矩阵传输至上位机,输入到2D-CNN模型进行故障诊断.在硬件方面,设计了电源电路和信号调理电路,实现了A/D 转换.测试结果表明:系统采集准确率大于95%,动态范围为97.45 dB,轴承故障诊断准确率大于95%,简化了2D-CNN轴承故障检测流程.
Abstract
In order to solve the problems of complex model deployment process and waste of resources at the acquisition end caused by encoding one-dimensional signals into two-dimensional feature matrices before 2D-CNN bearing fault diagnosis,a bearing fault diagnosis system based on the implementation of Gram's corner field edges was designed.The system took ZYNQ as the core,completed the signal preprocessing such as Gram angle field encoding through the PL(programmable logic)end,and used Ethernet to transmit the feature matrix to the host computer,which was input to the 2D-CNN model for fault diagnosis.In terms of hardware,the power supply circuit and signal conditioning circuit were designed to realize A/D conversion.The test results show that the system acquisition accuracy is greater than 95%,the dynamic range is 97.45 dB,the bearing fault diagnosis accuracy is greater than 95%,and the 2D-CNN bearing fault detection process is simplified.
关键词
格拉姆角场/轴承故障诊断/卷积神经网络/ZYNQ/信号预处理Key words
Gram angle field/bearing fault diagnosis/convolutional neural network/ZYNQ/signal preprocessing引用本文复制引用
出版年
2024
NETL
NSTL
万方数据