工程科学学报，第39卷，第5期：769-777.2017年5月 Chinese Journal of Engineering,Vol.39,No.5:769-777,May 2017 D0I:10.13374/j.issn2095-9389.2017.05.016;htp://journals.ustb.edu.cn 基于多域流形的行星齿轮箱局部故障识别 赵 川，冯志鹏区 北京科技大学机械工程学院，北京100083 ☒通信作者，E-mail:fengzp@ustb.cdu.cn 摘要行星齿轮箱振动信号包含多种频率成分和噪声干扰，频谱具有复杂的边带结构，容易对故障识别造成误导甚至引起 错判.在不同故障状态下，行星齿轮箱振动信号的多域特征量将偏离正常范围且偏离程度不同，根据这一特点，提取振动信 号的时域，频域特征参量用于故障识别.为了避免传统分析方法中负频率及虚假模态问题，增强对噪声干扰的鲁棒性，采用 局部均值分解法将信号自适应地分解为单分量之和，提取时频域单分量瞬时幅值能量.针对多域特征空间构造过程中出现 的高维及非线性问题，采用流形学习对数据进行降维处理.提出基于改进的虚假近邻点的本征维数估计及最优k邻域确定方 法，并通过等距映射对多域特征空间进行降维分析.对于行星齿轮箱实验信号，根据样本流形特征聚类结果，分别识别出了 太阳轮、行星轮和齿圈的局部故障，从而验证了上述方法的有效性 关键词行星齿轮箱；故障识别；局部均值分解；本征维数估计：多域流形 分类号TH17 Localized fault identification of planetary gearboxes based on multiple-domain manifold ZHAO Chuan,FENG Zhi-peng School of Mechanical Engineering,University of Science and Technology Beijing,Beijing 100083,China Corresponding author,E-mail:fengzp@ustb.edu.cn ABSTRACT The vibration signals of planetary gearboxes are composed of complex frequency components and interfering noises,and their spectra have intricate sidebands,which cause difficulty in and even misleading fault identification.In different fault cases,the vibration signatures in multiple domains typically differ from normal states with different discrepancies.Based on this hypothesis,time and frequency domain features are extracted for the purposes of fault identification.The vibration signal is adaptively decomposed into a set of mono-components,and the instantaneous energy of each mono-component is calculated in time-frequency domain by exploiting the merits of local mean decomposition,including its better robustness to noise and freedom from pseudo-mode and negative frequency problems.Manifold learning is utilized to tackle the high-dimensionality and non-linearity aspects of multiple-domain feature space construction.A new method is proposed for estimating the intrinsic dimension and selecting the k-nearest neighborhood based on the improved pseudo-nearest neighbor.In addition,isometric feature mapping (ISOMAP)is utilized to reduce the dimensions of the mul- tiple-domain feature space.The proposed method is validated by analyzing the planetary gearbox lab experimental dataset.Based on the clustering analysis results of the extracted manifold features,the localized faults on the sun,planet,and ring gears are successfully identified. KEY WORDS planetary gearboxes;fault identification;local mean decomposition;estimation of intrinsic dimension;multiple-do- main manifold 行星齿轮箱作为传动系统中的关键部件在风力发电机组中应用广泛.风力发电机组的工作环境造成行 收稿日期：2016-06-24 基金项目：国家自然科学基金资助项目(11272047,51475038)工程科学学报,第 39 卷,第 5 期:769鄄鄄777,2017 年 5 月 Chinese Journal of Engineering, Vol. 39, No. 5: 769鄄鄄777, May 2017 DOI: 10. 13374 / j. issn2095鄄鄄9389. 2017. 05. 016; http: / / journals. ustb. edu. cn 基于多域流形的行星齿轮箱局部故障识别 赵 川, 冯志鹏苣 北京科技大学机械工程学院, 北京 100083 苣 通信作者, E鄄mail: fengzp@ ustb. edu. cn 摘 要 行星齿轮箱振动信号包含多种频率成分和噪声干扰,频谱具有复杂的边带结构,容易对故障识别造成误导甚至引起 错判. 在不同故障状态下,行星齿轮箱振动信号的多域特征量将偏离正常范围且偏离程度不同,根据这一特点,提取振动信 号的时域、频域特征参量用于故障识别. 为了避免传统分析方法中负频率及虚假模态问题,增强对噪声干扰的鲁棒性,采用 局部均值分解法将信号自适应地分解为单分量之和,提取时频域单分量瞬时幅值能量. 针对多域特征空间构造过程中出现 的高维及非线性问题,采用流形学习对数据进行降维处理. 提出基于改进的虚假近邻点的本征维数估计及最优 k 邻域确定方 法,并通过等距映射对多域特征空间进行降维分析. 对于行星齿轮箱实验信号,根据样本流形特征聚类结果,分别识别出了 太阳轮、行星轮和齿圈的局部故障,从而验证了上述方法的有效性. 关键词 行星齿轮箱; 故障识别; 局部均值分解; 本征维数估计; 多域流形 分类号 TH17 收稿日期: 2016鄄鄄06鄄鄄24 基金项目: 国家自然科学基金资助项目 (11272047,51475038) Localized fault identification of planetary gearboxes based on multiple鄄domain manifold ZHAO Chuan, FENG Zhi鄄peng 苣 School of Mechanical Engineering, University of Science and Technology Beijing, Beijing 100083, China 苣 Corresponding author, E鄄mail: fengzp@ ustb. edu. cn ABSTRACT The vibration signals of planetary gearboxes are composed of complex frequency components and interfering noises, and their spectra have intricate sidebands, which cause difficulty in and even misleading fault identification. In different fault cases, the vibration signatures in multiple domains typically differ from normal states with different discrepancies. Based on this hypothesis, time and frequency domain features are extracted for the purposes of fault identification. The vibration signal is adaptively decomposed into a set of mono鄄components, and the instantaneous energy of each mono鄄component is calculated in time鄄frequency domain by exploiting the merits of local mean decomposition, including its better robustness to noise and freedom from pseudo鄄mode and negative frequency problems. Manifold learning is utilized to tackle the high鄄dimensionality and non鄄linearity aspects of multiple鄄domain feature space construction. A new method is proposed for estimating the intrinsic dimension and selecting the k鄄nearest neighborhood based on the improved pseudo鄄nearest neighbor. In addition, isometric feature mapping (ISOMAP) is utilized to reduce the dimensions of the mul鄄 tiple鄄domain feature space. The proposed method is validated by analyzing the planetary gearbox lab experimental dataset. Based on the clustering analysis results of the extracted manifold features, the localized faults on the sun, planet, and ring gears are successfully identified. KEY WORDS planetary gearboxes; fault identification; local mean decomposition; estimation of intrinsic dimension; multiple鄄do鄄 main manifold 行星齿轮箱作为传动系统中的关键部件在风力发 电机组中应用广泛. 风力发电机组的工作环境造成行