基于多域流形的行星齿轮箱局部故障识别

发布时间：2018-03-11 00:03

本文选题：行星齿轮箱　切入点：故障识别　出处：《工程科学学报》2017年05期 　论文类型：期刊论文

【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.
[Abstract]:The vibration signal of planetary gearbox contains many kinds of frequency components and noise interference, and the frequency spectrum has complex sideband structure, which is easy to mislead and even cause misjudgment to fault identification. The multi-domain characteristic of the planetary gearbox vibration signal will deviate from the normal range and the degree of deviation will be different. According to this characteristic, the time domain of vibration signal is extracted. In order to avoid the negative frequency and false modal problems in traditional analysis methods and enhance the robustness to noise interference, the local mean decomposition method is used to decompose the signal adaptively into the sum of single components. The instantaneous amplitude energy of single component in time-frequency domain is extracted. In order to solve the problem of high dimension and nonlinearity in the construction of multi-domain feature space, Using manifold learning to reduce the dimension of the data, the eigendimension estimation and the optimal k-neighborhood determination method based on improved false nearest neighbor points are proposed. For the experimental signals of planetary gearbox, the local faults of solar wheel, planetary gear and gear ring are identified according to the clustering results of sample manifold characteristics. The effectiveness of the above method is verified.
【作者单位】：北京科技大学机械工程学院;
【基金】：国家自然科学基金资助项目(11272047,51475038)
【分类号】：TH132.41

【相似文献】