论文部分内容阅读
摘要: 针对旋转机械不同故障可能分布于不同故障流形,提出了基于多故障流形的旋转机械故障诊断方法。该方法分别提取每一类故障对应的故障流形,并在多故障流形上进行新增样本的故障识别。针对所需解决的低维流形提取、流形内蕴维数选取和多故障流形上的故障识别问题,分别采用线性局部切空间排列算法和免疫遗传算法来进行低维故障流形提取和流形内蕴维数选取,并通过故障样本重构误差这一新的判别准则进行故障识别。齿轮箱故障模拟实验的结果验证了此方法的有效性。关键词: 故障诊断; 旋转机械; 多故障流形; 局部切空间排列算法
中图分类号:TH165+.3; TN911.2文献标志码: A文章编号: 10044523(2015)02030907
DOI:10.16385/j.cnki.issn.10044523.2015.02.018
引言
在对旋转机械进行故障诊断时,为了获取尽可能多的故障信息来对故障进行更加全面、综合地描述,通常都需要从时域、频域以及时频域等多方面提取大量的故障特征。然而,各故障特征间通常都具有不同程度的耦合关系,这些冗余信息影响了故障诊断的效果[1]。
流形学习是一种新型非线性特征融合方法,能够有效地提取出嵌入在高维观察空间中的低维流形结构[2,3]。然而现有的基于流形学习的故障诊断方法都是将各类故障映射到一个低维流形上进行故障诊断,即假定所有故障都分布于同一个流形之上[1~5]。虽然文献[6]提出了基于多流形分析的故障诊断,但该方法只是通过由振动信号相空间重构得到的空间流形提取故障特征向量,并输入多路主成分分析算法进行二次故障特征提取,其本质仍然是将各类故障样本映射到一个流形上进行故障诊断。由于旋转机械的故障机理十分复杂,不同零部件的故障、同一零部件的不同故障以及故障的位置都将影响故障的表现形式,因此并非所有故障都一定分布于同一流形,且不同故障对应的故障流形的内蕴维数也可能各不相同,将所有故障样本都映射到一个低维故障流形的做法势必将影响故障诊断精度的进一步提高。
文献[7]首次提出在多流形上来进行面部表情识别,并取得了比传统的基于单一流形的面部表情识别更好的识别效果。本文则在一类故障对应于一个故障流形的假设之上,将基于多流形的模式识别方法引入到故障诊断中,提出基于多故障流形的旋转机械故障诊断方法。该方法分别提取每一类故障对应的故障流形,并在多故障流形上对新增样本进行故障识别。
1基于多故障流形的旋转机械故障诊断方法原理
1.1基于多故障流形的旋转机械故障诊断算法构架现有的基于流形学习的故障诊断方法都是将各类故障样本映射到一个低维故障流形之上进行故障诊断,如图1所示。
然而实际情况下并非所有故障都一定分布于同一个流形之上,将所有故障样本都映射到一个低维故障流形的做法势必将阻碍故障诊断精度的进一步提高,因此更为行之有效的故障流形提取方法是分别提取每一类故障对应的故障流形。本文提出的基于多故障流形的故障诊断方法,其基本思想是认为各类故障分布于不同的故障流形之上,并分别提取每一类故障对应的故障流形,且优化选取各故障流形对应的内蕴维数,最终在多故障流形之上完成故障诊断,基于多故障流形的旋转机械故障诊断方法的算法构架如图2所示。
结果可以看出,采用EMD进行预处理后得到的故障诊断精度更高。这是由于故障振动信号一般都是非平稳、非线性的,且包含的频率成分通常都比较复杂,因此难以从原始故障振动信号中提取出可辨识性高的故障特征。EMD作为一种时频信号分析方法,能够将故障振动信号分解为一系列包含不同频带的IMF分量,从这些IMF分量中能够有效地提取出具有更高可辨识性的故障特征。同时,由于单域特征所包含的故障信息有限,无法全面有效地对故障进行描述,因此,仅仅采用时域特征或者频域特征来进行故障诊断,得到的结果都不够理想。而由时域特征和频域特征组成的混合域特征集从时域和频域两个方面对故障进行了更加全面、综合地描述,提供了更加丰富的故障信息,因此采用混合域特征集进行故障诊断的精度显然更高。
4结论
本文提出基于多故障流形的旋转机械故障诊断方法,根据实验结果以及结果分析可以得出以下结论:(1)混合域特征集可有效地表征故障的特性,且采用EMD对原始振动信号进行预处理可以提高所提取的故障特征的可辨识性;(2)故障特征间存在的大量冗余信息削弱了故障特征的可辨识性,而采用维数约简方法对原始高维故障特征集进行特征融合可以消除特征集中的冗余信息,提高了故障诊断的精度;(3)采用多故障流形的故障诊断方法的效果优于现有的基于单一故障流形的故障诊断方法,同时采用LLTSA进行低维故障流形提取的效果优于LPP;(4)齿轮箱故障模拟实验的结果验证了本文方法的有效性,由实验结果可以看出采用多故障流形的故障诊断精度为97.67%,而基于单一故障流形的故障诊断精度为87%。
本文的后续研究可以从以下两个方面进行展开:(1)针对复合故障的识别展开深入的研究,即识别出复合故障中所包含的故障类别; (2)本文目前采用的是故障样本重构误差来进行故障识别,后续研究可以针对多故障流形上的故障识别方法展开深入的研究。
参考文献:
[1]Li Feng, Tang Baoping, Yang Rongsong. Rotating machine fault diagnosis using dimension reduction with linear local tangent space alignmet[J]. Measurement, 2013,(46):2 525—2 539.
[2]Zhang Z Y, Zha H Y. Principal manifolds and nonlinear dimensionality reduction via tangent space alignment[J]. SIAM Journal on Scientific Computing, 2005,8(4):406—424. [3]Rweis S T, Saul L K, Nonlinear dimensionality reduction by locally linear embedding[J]. Science, 2000,(290):2 323—2 326.
[4]Li B W, Zhang Y. Supervised locally linear embedding projection for machinery fault diagnosis[J]. Mechanical Systems and Signal Processing, 2011,(25):3 125—3 134.
[5]Tang Baoping, Song Tao, Li Feng, et al. Fault diagnosis for wind turbine transmission system based on manifold learning and Shannon wavelet support vector machine[J]. Renewable Energy, 2014,(62):1—9.
[6]Li Min, Xu Jinwu, Yang Jianhong, et al. Multiple manifolds analysis and its application to fault diagnosis[J]. Mechanical Systems and Signal Processing, 2009,(23):2 500—2 509.
[7]Xiao Rui, Zhao Qijun, Zhang David. Facial expression recognition on multiple manifods[J]. Pattern Recognition, 2011,(44):107—116.
[8]Tianhao Zhang, Jie Yang, Deli Zhao, et al. Linear local tangent space alignment and application to face recognition [J]. Mechanical Systems and Signal Processing, 2007,(70):1 547—1 553.
[9]Luo Jiawei, Wang Ting. Motif discovery using an immune genetic algorithm[J]. Journal of Theoretical Biology, 2010,(264):319—325.
[10]He Xiaofei, Yan Shuichen, Hu Yutao, et al. Face recognitions using Laplacian faces[J]. IEEE Tramsactions on Pattern Analysis and Machine Intelligence,2005,27(3):328—340.
[11]Jolliffe I T. Principle Component Analysis[M]. New York: Springer, 1986.
[12]Huang N E. The empirical mode decomposition and the Hilbert spectrum for nonlinear and nonstationary time series analysis[J]. Royal Society of London Proceedings Series A,1998,454(1971):903—995.
[13]Shen Z J, Chen X F, Zhang X L, et al. A novel intelligent gear fault diagnosis model based on EMD and multiclass TSVM[J]. Measurement, 2012,(45):30—40.
[14]Lei Y G, He Z J, Zi Y Y. A new approach to intelligent fault diagnosis of rotating machinery[J]. Expert Systems with Application, 2008,(35):1 593—1 600.
Abstract: The existing fault diagnosis methods based on manifold learning assume that all the faults distribute on a single manifold, however the faults may distribute on different manifolds in practical applications. Aiming at this problem, rotating machinery fault diagnosis method based on multiple fault manifolds is proposed. Firstly, mixeddomain features are extracted from the vibration signals to characterize the property of the faults, and the vibration signals are also preprocessed by empirical model decomposition before feature extraction. Then, the corresponding fault manifold of each fault is extracted from the highdimensional fault samples. In the method, linear local tangent space alignment is applied to solve the problem of lowdimensional manifold extraction, and immune genetic algorithm is used to select the intrinsic dimensionality of fault manifold. At last, the test samples are respectively projected to all the fault manifolds, and the projection errors are used as the criterion to determine the fault types of the test samples. In order to verify the effectiveness of the proposed fault diagnosis method, the method is applied to diagnose the faults of the gear box. The experimental results indicate that feature compression can remove the redundant information between features, and moreover fault diagnosis method based on multiple fault manifolds can obtain even better performance than those methods which project all the faults to a single lowdimensional manifold.
Key words: fault diagnosis; rotating machinery; multiple fault manifolds; linear local tangent space alignment
中图分类号:TH165+.3; TN911.2文献标志码: A文章编号: 10044523(2015)02030907
DOI:10.16385/j.cnki.issn.10044523.2015.02.018
引言
在对旋转机械进行故障诊断时,为了获取尽可能多的故障信息来对故障进行更加全面、综合地描述,通常都需要从时域、频域以及时频域等多方面提取大量的故障特征。然而,各故障特征间通常都具有不同程度的耦合关系,这些冗余信息影响了故障诊断的效果[1]。
流形学习是一种新型非线性特征融合方法,能够有效地提取出嵌入在高维观察空间中的低维流形结构[2,3]。然而现有的基于流形学习的故障诊断方法都是将各类故障映射到一个低维流形上进行故障诊断,即假定所有故障都分布于同一个流形之上[1~5]。虽然文献[6]提出了基于多流形分析的故障诊断,但该方法只是通过由振动信号相空间重构得到的空间流形提取故障特征向量,并输入多路主成分分析算法进行二次故障特征提取,其本质仍然是将各类故障样本映射到一个流形上进行故障诊断。由于旋转机械的故障机理十分复杂,不同零部件的故障、同一零部件的不同故障以及故障的位置都将影响故障的表现形式,因此并非所有故障都一定分布于同一流形,且不同故障对应的故障流形的内蕴维数也可能各不相同,将所有故障样本都映射到一个低维故障流形的做法势必将影响故障诊断精度的进一步提高。
文献[7]首次提出在多流形上来进行面部表情识别,并取得了比传统的基于单一流形的面部表情识别更好的识别效果。本文则在一类故障对应于一个故障流形的假设之上,将基于多流形的模式识别方法引入到故障诊断中,提出基于多故障流形的旋转机械故障诊断方法。该方法分别提取每一类故障对应的故障流形,并在多故障流形上对新增样本进行故障识别。
1基于多故障流形的旋转机械故障诊断方法原理
1.1基于多故障流形的旋转机械故障诊断算法构架现有的基于流形学习的故障诊断方法都是将各类故障样本映射到一个低维故障流形之上进行故障诊断,如图1所示。
然而实际情况下并非所有故障都一定分布于同一个流形之上,将所有故障样本都映射到一个低维故障流形的做法势必将阻碍故障诊断精度的进一步提高,因此更为行之有效的故障流形提取方法是分别提取每一类故障对应的故障流形。本文提出的基于多故障流形的故障诊断方法,其基本思想是认为各类故障分布于不同的故障流形之上,并分别提取每一类故障对应的故障流形,且优化选取各故障流形对应的内蕴维数,最终在多故障流形之上完成故障诊断,基于多故障流形的旋转机械故障诊断方法的算法构架如图2所示。
结果可以看出,采用EMD进行预处理后得到的故障诊断精度更高。这是由于故障振动信号一般都是非平稳、非线性的,且包含的频率成分通常都比较复杂,因此难以从原始故障振动信号中提取出可辨识性高的故障特征。EMD作为一种时频信号分析方法,能够将故障振动信号分解为一系列包含不同频带的IMF分量,从这些IMF分量中能够有效地提取出具有更高可辨识性的故障特征。同时,由于单域特征所包含的故障信息有限,无法全面有效地对故障进行描述,因此,仅仅采用时域特征或者频域特征来进行故障诊断,得到的结果都不够理想。而由时域特征和频域特征组成的混合域特征集从时域和频域两个方面对故障进行了更加全面、综合地描述,提供了更加丰富的故障信息,因此采用混合域特征集进行故障诊断的精度显然更高。
4结论
本文提出基于多故障流形的旋转机械故障诊断方法,根据实验结果以及结果分析可以得出以下结论:(1)混合域特征集可有效地表征故障的特性,且采用EMD对原始振动信号进行预处理可以提高所提取的故障特征的可辨识性;(2)故障特征间存在的大量冗余信息削弱了故障特征的可辨识性,而采用维数约简方法对原始高维故障特征集进行特征融合可以消除特征集中的冗余信息,提高了故障诊断的精度;(3)采用多故障流形的故障诊断方法的效果优于现有的基于单一故障流形的故障诊断方法,同时采用LLTSA进行低维故障流形提取的效果优于LPP;(4)齿轮箱故障模拟实验的结果验证了本文方法的有效性,由实验结果可以看出采用多故障流形的故障诊断精度为97.67%,而基于单一故障流形的故障诊断精度为87%。
本文的后续研究可以从以下两个方面进行展开:(1)针对复合故障的识别展开深入的研究,即识别出复合故障中所包含的故障类别; (2)本文目前采用的是故障样本重构误差来进行故障识别,后续研究可以针对多故障流形上的故障识别方法展开深入的研究。
参考文献:
[1]Li Feng, Tang Baoping, Yang Rongsong. Rotating machine fault diagnosis using dimension reduction with linear local tangent space alignmet[J]. Measurement, 2013,(46):2 525—2 539.
[2]Zhang Z Y, Zha H Y. Principal manifolds and nonlinear dimensionality reduction via tangent space alignment[J]. SIAM Journal on Scientific Computing, 2005,8(4):406—424. [3]Rweis S T, Saul L K, Nonlinear dimensionality reduction by locally linear embedding[J]. Science, 2000,(290):2 323—2 326.
[4]Li B W, Zhang Y. Supervised locally linear embedding projection for machinery fault diagnosis[J]. Mechanical Systems and Signal Processing, 2011,(25):3 125—3 134.
[5]Tang Baoping, Song Tao, Li Feng, et al. Fault diagnosis for wind turbine transmission system based on manifold learning and Shannon wavelet support vector machine[J]. Renewable Energy, 2014,(62):1—9.
[6]Li Min, Xu Jinwu, Yang Jianhong, et al. Multiple manifolds analysis and its application to fault diagnosis[J]. Mechanical Systems and Signal Processing, 2009,(23):2 500—2 509.
[7]Xiao Rui, Zhao Qijun, Zhang David. Facial expression recognition on multiple manifods[J]. Pattern Recognition, 2011,(44):107—116.
[8]Tianhao Zhang, Jie Yang, Deli Zhao, et al. Linear local tangent space alignment and application to face recognition [J]. Mechanical Systems and Signal Processing, 2007,(70):1 547—1 553.
[9]Luo Jiawei, Wang Ting. Motif discovery using an immune genetic algorithm[J]. Journal of Theoretical Biology, 2010,(264):319—325.
[10]He Xiaofei, Yan Shuichen, Hu Yutao, et al. Face recognitions using Laplacian faces[J]. IEEE Tramsactions on Pattern Analysis and Machine Intelligence,2005,27(3):328—340.
[11]Jolliffe I T. Principle Component Analysis[M]. New York: Springer, 1986.
[12]Huang N E. The empirical mode decomposition and the Hilbert spectrum for nonlinear and nonstationary time series analysis[J]. Royal Society of London Proceedings Series A,1998,454(1971):903—995.
[13]Shen Z J, Chen X F, Zhang X L, et al. A novel intelligent gear fault diagnosis model based on EMD and multiclass TSVM[J]. Measurement, 2012,(45):30—40.
[14]Lei Y G, He Z J, Zi Y Y. A new approach to intelligent fault diagnosis of rotating machinery[J]. Expert Systems with Application, 2008,(35):1 593—1 600.
Abstract: The existing fault diagnosis methods based on manifold learning assume that all the faults distribute on a single manifold, however the faults may distribute on different manifolds in practical applications. Aiming at this problem, rotating machinery fault diagnosis method based on multiple fault manifolds is proposed. Firstly, mixeddomain features are extracted from the vibration signals to characterize the property of the faults, and the vibration signals are also preprocessed by empirical model decomposition before feature extraction. Then, the corresponding fault manifold of each fault is extracted from the highdimensional fault samples. In the method, linear local tangent space alignment is applied to solve the problem of lowdimensional manifold extraction, and immune genetic algorithm is used to select the intrinsic dimensionality of fault manifold. At last, the test samples are respectively projected to all the fault manifolds, and the projection errors are used as the criterion to determine the fault types of the test samples. In order to verify the effectiveness of the proposed fault diagnosis method, the method is applied to diagnose the faults of the gear box. The experimental results indicate that feature compression can remove the redundant information between features, and moreover fault diagnosis method based on multiple fault manifolds can obtain even better performance than those methods which project all the faults to a single lowdimensional manifold.
Key words: fault diagnosis; rotating machinery; multiple fault manifolds; linear local tangent space alignment