分拣和包装设备运行状态实时监测及故障诊断技术研究

孙奕; 宋睿; 靳威; 左右宇; 舒晔; 李嘉杰

分拣和包装设备运行状态实时监测及故障诊断技术研究

引用本文：孙奕，宋睿，靳威，左右宇，舒晔，李嘉杰.分拣和包装设备运行状态实时监测及故障诊断技术研究[J].计算技术与自动化,2026,(2):215-220

摘要点击次数: 36

全文下载次数: 0

作者	单位
孙奕，宋睿，靳威，左右宇，舒晔，李嘉杰	（广东电网公司计量中心,广东广州 510030）

中文摘要:针对分拣和包装设备故障诊断技术正确率低、难以应用于实际的问题，设计了一种新型分拣和包装设备实时监测和故障诊断方法。这种分拣和包装设备，通过数据采集卡采集设备运行产生的振动信号，利用遗传算法对变分模态分解进行优化，提高变分模态分解的分解效率和准确性；通过变分模态分解分解振动信号来提取信号特征、对信号进行实时分析，完成实时监测；通过改进型深度置信网络模型对设备振动信号进行训练，根据变分模态分解提取出的故障特征，完成设备的故障诊断；通过设计的诊断方法与文献［4］方法的实验对比结果，得到本方法的正确率为98.17%，文献［4］方法的正确率为96.94%，验证了该故障诊断方法的优越性。

中文关键词:实时监测故障诊断改进型深度置信网络变分模态分解

Research on Real-time Monitoring and Fault Diagnosis Technology for Sorting and Packaging Equipment Operation Status

Abstract:For the problems of low correct rate and difficult to be applied in practice of sorting and packaging equipment fault diagnosis technology, a new type of sorting and packaging equipment real-time monitoring and fault diagnosis method is designed. In this paper, a sorting and packaging equipment is designed to collect the vibration signals generated by the operation of the equipment through the data acquisition card, optimize the variational pattern decomposition by using genetic algorithm to improve the decomposition accuracy and efficiency of the variational pattern decomposition; extract the signal characteristics and analyze the signals in real time through the decomposition of vibration signals by the variational pattern decomposition, and complete the real-time monitoring; train the vibration signals of the equipment through the improved deep confidence network model; and extract the vibration signals from the vibration signals based on the variational pattern decomposition, and analyze the vibration signals in real time through the improved deep confidence network model. Signal training through the improved deep confidence network model on the equipment voltage signal, according to the fault features extracted by the variational modal decomposition, to finish the fault diagnosis of the equipment; through the diagnostic approach designed in this paper and the experimental comparison of the literature four ways to get the right rate of this paper's method is 98.17%, the correct rate of the literature four ways to get the right rate of 96.94%, which verified this paper's fault diagnostic method of the superiority of the method.

keywords:real-time monitoring fault diagnosis improved deep confidence network variational modal decomposition

查看全文 查看/发表评论 下载pdf阅读器