针对滚动轴承尤其是变工况条件下很难或无法获取大量带标签的振动数据,以致诊断准确率低的问题,提出一种基于变分模态分解(variationalmodedecomposition,VMD)及多特征构造和迁移学习相结合的滚动轴承故障诊断方法。该方法利用VMD对滚...针对滚动轴承尤其是变工况条件下很难或无法获取大量带标签的振动数据,以致诊断准确率低的问题,提出一种基于变分模态分解(variationalmodedecomposition,VMD)及多特征构造和迁移学习相结合的滚动轴承故障诊断方法。该方法利用VMD对滚动轴承各状态振动信号进行分解,得到一系列固有模态函数,对其构成的矩阵进行奇异值分解求奇异值及奇异值熵,再结合振动信号的时域、频域特征构造多特征集。同时引入半监督迁移成分分析方法(semisupervised transfer component analysis,SSTCA),并对其核函数进行多核构造,将不同工况样本特征共同映射到一个共享再生核Hilbert空间,进而提高数据类内紧凑性和类间区分性。采用最大均值差异嵌入法选择更有效的数据作为源域,将源域特征样本输入支持向量机(supportvectormachine,SVM)进行训练,测试映射后的目标域特征样本。实验表明,所提多核SSTCA-SVM方法与其他方法相比较,在变工况下滚动轴承多状态分类中具有更高准确率。展开更多
The axial piston pump usually works under variable speed conditions.It is important to evaluate the health status of the axial piston pump under the variable speed condition.Aiming at the characteristic signals obtain...The axial piston pump usually works under variable speed conditions.It is important to evaluate the health status of the axial piston pump under the variable speed condition.Aiming at the characteristic signals obtained under different wear levels of the port plate,a feature signal extraction method under variable speed conditions is proposed.Firstly,the combination of complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)energy spectrum and fast spectral kurtosis principle is used to accurately extract the intrinsic mode function(IMF)component containing the sensitive information of the degraded feature.Then,the aspect ratio analysis method of the angle domain variational mode decomposition(VMD)is used to process the feature index containing the sensitive information of the degraded feature.In order to evaluate the health status of the axial piston pump under variable speed,the vibration reliability analysis method for axial piston pump based on Weibull proportional failure rate model is proposed.The experimental results show that the proposed method can accurately evaluate the health status of the axial piston pump.展开更多
文摘针对滚动轴承尤其是变工况条件下很难或无法获取大量带标签的振动数据,以致诊断准确率低的问题,提出一种基于变分模态分解(variationalmodedecomposition,VMD)及多特征构造和迁移学习相结合的滚动轴承故障诊断方法。该方法利用VMD对滚动轴承各状态振动信号进行分解,得到一系列固有模态函数,对其构成的矩阵进行奇异值分解求奇异值及奇异值熵,再结合振动信号的时域、频域特征构造多特征集。同时引入半监督迁移成分分析方法(semisupervised transfer component analysis,SSTCA),并对其核函数进行多核构造,将不同工况样本特征共同映射到一个共享再生核Hilbert空间,进而提高数据类内紧凑性和类间区分性。采用最大均值差异嵌入法选择更有效的数据作为源域,将源域特征样本输入支持向量机(supportvectormachine,SVM)进行训练,测试映射后的目标域特征样本。实验表明,所提多核SSTCA-SVM方法与其他方法相比较,在变工况下滚动轴承多状态分类中具有更高准确率。
文摘为了有效检测轨道波磨故障,提出一种基于参数优化变分模态分解(VMD,VariableMode Decomposition)和平滑伪维格纳分布(SPWVD,SmoothPseudo Wigner VilleDistribution)的轨道波磨辨识方法。采用变步长最小均方(VSSLMS,Variable Step Size Least Mean Square)算法对列车轴箱振动加速度原始信号滤波;对滤波后的信号进行变分模态分解,将分解信号包络熵作为轨道波磨辨识的指标;采用平滑伪维格纳分布对分解后的信号进行时频分析,确定波磨发生的位置及波长;通过仿真信号与实例验证方法的有效性。验证结果表明,该方法可提高轨道波磨辨识的准确性,辅助轨道维修和养护。
基金Supported by the National Key Research and Development Program of China(No.2019YFB2005204)the National Natural Science Foundation of China(No.52075469,51675461,11673040)+1 种基金the Key Research and Development Program of Hebei Province(No.19273708D)the Open Foundation of the State Key Laboratory of Fluid Power and Mechatronic Systems(No.GZKF-201922).
文摘The axial piston pump usually works under variable speed conditions.It is important to evaluate the health status of the axial piston pump under the variable speed condition.Aiming at the characteristic signals obtained under different wear levels of the port plate,a feature signal extraction method under variable speed conditions is proposed.Firstly,the combination of complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)energy spectrum and fast spectral kurtosis principle is used to accurately extract the intrinsic mode function(IMF)component containing the sensitive information of the degraded feature.Then,the aspect ratio analysis method of the angle domain variational mode decomposition(VMD)is used to process the feature index containing the sensitive information of the degraded feature.In order to evaluate the health status of the axial piston pump under variable speed,the vibration reliability analysis method for axial piston pump based on Weibull proportional failure rate model is proposed.The experimental results show that the proposed method can accurately evaluate the health status of the axial piston pump.
