首页|一种旋转机械综合故障检测和模式识别模型

一种旋转机械综合故障检测和模式识别模型

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
针对传统故障诊断方法只关注故障检测部分,而对样本是否存在故障的研究较少的问题,提出了一种基于自适应噪声完备经验模态分解(CEEMDAN)-注意熵(AE)和黏菌算法优化极限学习机(SMA-ELM)的旋转机械综合故障诊断模型.首先,针对正常样本和故障样本的复杂性差异,建立了注意熵阈值,计算旋转机械的 AE,并将其与阈值进行了比较,若熵值小于该阈值则表明样本存在故障,反之样本是健康的;然后,利用 CEEMDAN对故障样本的振动信号进行了分解,提取前 6 阶分量的 AE值;最后,将故障特征输入至 SMA-ELM 模型中进行了故障识别,利用 3 种旋转机械故障数据集对该综合故障诊断模型的可靠性进行了研究.研究结果表明:该阈值设置方法可以 100％准确地检测样本是否存在故障,后续的故障诊断模型能够准确地检测出样本的故障类型,识别准确率分别达到了 99.44％、100％和 98％.该综合故障诊断模型能够避免正常样本被误判为故障样本,为旋转机械的故障检测提供了一种可行的思路.
A comprehensive fault detection and pattern recognition model for rotating machinery
Aiming at the problem that the traditional fault diagnosis methods only focus on the fault detection part and there are few studies on whether there are faults in the sample.A comprehensive fault diagnosis model for rotating machinery based on complementary ensemble empirical mode decomposition with adaptive noise(CEEMDAN)-attention entropy(AE)and slime mould algorithm optimized extreme learning machine(SMA-ELM)was proposed.Firstly,according to the complexity difference between normal samples and fault samples,the attention entropy threshold was established.The AE value of the rotating machine was calculated and compared with the threshold value.If the entropy value was less than the threshold value,it was indicated that the sample had a fault,otherwise,the sample was healthy.Then,CEEMDAN was used to decompose the vibration signal of the fault sample,and extract the AE of the first 6 order components.Finally,the fault features were input into the SMA-ELM model for fault identification.The reliability of the comprehensive fault diagnosis model was studied using three rotating machinery fault datasets.The research results show that the threshold setting method can detect whether there is a fault in the sample with 100％accuracy,and the subsequent fault diagnosis model can accurately detect the fault type of the sample with the recognition accuracy of 99.44％,100％and 98％,respectively.This comprehensive fault diagnosis model can prevent normal samples from being misjudged as faulty samples,providing a feasible approach for fault detection of rotating machinery.

rotating machineryrolling bearing comprehensive fault diagnosisfault thresholdattention entropy(AE)complementary ensemble empirical mode decomposition with adaptive noise(CEEMDAN)slime mould algorithm optimized extreme learning machine(SMA-ELM)

曹丽芳、袁征、尹久、郭海涛

展开 >

黄河水利职业技术学院,河南 开封 475004

湖北轻工职业技术学院 机电工程学院,湖北 武汉 430070

华南理工大学 土木交通学院,广东 广州 510640

旋转机械 滚动轴承综合故障诊断 故障阈值 注意熵 自适应噪声完备经验模态分解 黏菌算法优化极限学习机

河南省科技攻关项目江苏省现代农机装备与技术示范推广项目

222102220115NJ2019-06

2024

10.3969/j.issn.1001-4551.2024.08.007

机电工程
浙江大学　浙江省机电集团有限公司

机电工程

CSTPCD北大核心
影响因子：0.785
ISSN：1001-4551
年,卷(期)：2024.41(8)