基于张量奇异谱分解的机械故障特征提取方法研究

发布时间：2018-07-08 09:52

本文选题：张量分解 + 标准分解　；参考：《机械设计与制造》2017年07期

【摘要】：提出了将张量奇异谱分解运用于机械故障诊断特征提取,张量奇异谱分解(TSSA)是传统奇异谱分解(SVD)的扩展。由于传统奇异谱分解在处理非平稳、非线性的信号效果不理想,故障特征不明显。因此将传统的奇异谱分解延伸到三阶张量分解中,从而将一维的时间序列转换成为三阶的张量,然后运用标准(PARAFAC)张量分解,标准(PARAFAC)张量分解是把秩为R的张量分解为R个秩-1的张量的和,分解出原始张量的因子矩阵和权重,并重构回一维信号进而对信号的时域和频域做出分析。为了证明方法的有效性,将该方法应用于轴承故障信号的特征提取中,分别运用了仿真和实测信号做了分析,结果表明该方法不仅能有效地抑制噪声,明显地提取轴承故障信号特征,而且其效果要优于传统的奇异谱分解方法,具有一定的工程实践价值。
[Abstract]:In this paper, Zhang Liang singular spectral decomposition is applied to feature extraction of mechanical fault diagnosis. Zhang Liang singular spectral decomposition is an extension of traditional singular spectral decomposition. Because of the traditional singular spectrum decomposition in dealing with non-stationary, nonlinear signal effect is not ideal, fault characteristics are not obvious. Therefore, the traditional singular spectral decomposition is extended to the third-order Zhang Liang decomposition, and then the one-dimensional time series is transformed into the third-order Zhang Liang, and then the standard (PARAFAC) Zhang Liang decomposition is used. The standard (PARAFAC) Zhang Liang decomposition is to decompose the Zhang Liang with rank R into the sum of Zhang Liang with rank R, decompose the factor matrix and weight of the original Zhang Liang, and reconstruct the one-dimensional signal to analyze the time domain and frequency domain of the signal. In order to prove the validity of the method, the method is applied to the feature extraction of bearing fault signal. The simulation and the measured signal are used to analyze the method. The results show that the method can not only suppress the noise effectively. The feature of bearing fault signal is obviously extracted, and its effect is better than the traditional singular spectrum decomposition method, which has certain engineering practice value.
【作者单位】：武汉科技大学机械自动化学院;武钢股份设备管理部;
【基金】：国家自然科学基金(51475339) 武汉科技大学冶金装备及其控制教育部重点实验室开放基金资助项目(2015B11)
【分类号】：TH17

【相似文献】