首页|基于通道信息不平衡度的多元经验模态分解方法

基于通道信息不平衡度的多元经验模态分解方法

扫码查看
原文链接
  • 万方数据
  • 维普
在轴承多通道振动信号中,由于多通道故障信息不平衡会导致轴承故障诊断精度降低.针对这一问题,提出了一种考虑多通道故障信息不平衡的,基于多元经验模态分解(MEMD)的轴承故障特征提取与诊断方法.首先,分析了传统MEMD随机选择映射方向的缺陷,设计了依据通道间故障信息不平衡度自适应调整映射方向的策略,使分量信号中包含更多故障信息,并基于多元模态分解结果构造了特征空间;然后,基于冗余属性投影法对多通道提取的故障特征进行了融合,得到了多通道融合的本质故障特征;最后,采用反向传播(BP)神经网络进行了故障模式识别,设计了三层神经网络结构,且使用误差反向传播法进行了参数训练,并制定了基于BP神经网络的轴承故障诊断流程.研究结果表明:改进MEMD提取特征的类边界比传统方法更加明确,说明改进方法能够提取更具代表性的故障特征;从诊断精度看,与传统多元模态分解方法、完备集成辛几何分解方法相比,改进MEMD方法的诊断准确率最高,达到了99.5%.实验结果验证了改进方法在多通道故障诊断中是可行的,且从诊断精度上看,其具有一定的先进性.
Multivariate empirical mode decomposition method based on unbalanced channel information
In order to address the problem of low diagnostic accuracy caused by imbalanced fault information in multichannel vibration signals of bearings,a feature extraction and diagnosis method considering multichannel information imbalanced multivariate empirical mode decomposition(MEMD)was proposed.First,the shortcomings of traditional multivariate empirical mode decomposition in randomly selecting the mapping direction were analyzed,and a strategy was designed to adaptively adjust the mapping direction based on the imbalance of fault information between channels,so that the component signal contained more fault information,and a feature space based on the results of multivariate mode decomposition was constructed.Then,based on redundant attribute projection method,the fault features extracted from multiple channels were fused to obtain the essential fault features of multichannel fusion.Finally,the back propagation(BP)neural network was used for fault pattern recognition,a three-layer neural network structure was designed,and error back propagation method was used for parameter training.A bearing fault diagnosis process based on BP neural network was developed.The research result shows that the improved MEMD has a clearer class boundary for feature extraction compared to traditional method,indicating that the improved method can extract more representative fault features.From the perspective of diagnostic accuracy,comparing with the traditional multimodal decomposition and complete integrated symplectic geometry decomposition,the improved method possesses the highest diagnostic accuracy of 99.5%.It is verified by the experimental result that the improved method is feasible in multichannel fault diagnosis,and possesses a certain progressiveness in terms of diagnosis accuracy.

feature extraction and diagnosis of bearing faultsmultichannel samplinginformation imbalancemultivariate empirical mode decomposition(MEMD)redundant attribute projectionback propagation (BP) neural networkfeature space constructionessential fault c

陆春元、焦洪宇、卜王辉

展开 >

苏州市职业大学 机电工程学院,江苏 苏州 215104

常熟理工学院 汽车工程学院,江苏 苏州 215500

同济大学 机械与能源工程学院,上海 200092

轴承故障特征提取与诊断 多通道采样 信息不平衡 多元经验模态分解 冗余属性投影 反向传播(BP)神经网络 特征空间构造 本质故障特征

国家自然科学基金资助项目江苏政府留学奖学金资助项目

51605046JS-2017-188

2024

10.3969/j.issn.1001-4551.2024.02.011

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(2)
  • 23