基于相空间重构理论的滚动轴承故障诊断研究

发布时间：2018-06-21 05:04

本文选题：相空间重构 + 形态滤波　；参考：《武汉科技大学》2014年硕士论文

【摘要】：滚动轴承是各种旋转机械中应用最广泛的一种通用机械部件，它的运行状态是否正常往往影响整台机器的性能。因此，对滚动轴承进行故障诊断具有重要的意义。滚动轴承的故障诊断一般是对非线性时间序列表示的信号进行分析，比如特征提取、状态识别。主要研究内容如下： (1)滚动轴承信号往往含有噪声，为了降低噪声对特征提取的影响，因此有必要在特征提取之前对信号作降噪处理。本文提出了基于相空间重构技术的主分量分析降噪算法，并用仿真信号和滚动轴承故障实验数据证明了该方法在降噪方面的有效性。 (2)研究了形态滤波与基于相空间重构技术的主分量分析降噪算法相结合的特征提取算法。信号经基于相空间重构的主分量分析降噪方法降噪之后，再用形态滤波进行特征提取。仿真研究与滚动轴承故障内圈和外圈实验数据的实例分析，证明了该方法的有效性。 (3)研究了局部均值分解（LMD）与基于相空间重构技术的主分量分析降噪算法相结合的特征提取算法。信号基于相空间重构的主分量分析降噪方法降噪之后，再用LMD对其进行分解，选取能量最高的PF1进行包络谱分析。通过仿真试验和滚动轴承故障实验，结果表明该方法能够有效地提取出信号的故障特征。 (4)研究了多尺度排列熵与支持向量机结合的滚动轴承状态识别算法。通过计算各个尺度下滚动轴承四种状态信号的排列熵值，选择合适的尺度来构建特征向量，选取一定数量的特征向量样本并运用支持向量机分类器来对其进行分类，结果表明该方法对滚动轴承的正常、内圈故障、外圈故障、滚动体故障这四种状态具有很高的识别率。
[Abstract]:Rolling bearing is one of the most widely used universal mechanical parts in all kinds of rotating machinery. Whether its running state is normal or not often affects the performance of the whole machine. Therefore, the rolling bearing fault diagnosis is of great significance. The fault diagnosis of rolling bearings is usually based on the analysis of nonlinear time series signals, such as feature extraction and state recognition. The main research contents are as follows: (1) Rolling bearing signals often contain noise. In order to reduce the influence of noise on feature extraction, it is necessary to do noise reduction before feature extraction. In this paper, a principal component analysis (PCA) denoising algorithm based on phase space reconstruction is proposed. Simulation signals and rolling bearing fault data are used to prove the effectiveness of this method in noise reduction. The feature extraction algorithm based on morphological filtering and principal component analysis (PCA) de-noising algorithm based on phase space reconstruction is studied. After the signal is de-noised by principal component analysis (PCA) based on phase space reconstruction, morphological filtering is used for feature extraction. Simulation study and analysis of the experimental data of the inner ring and outer ring of rolling bearing fault, The validity of this method is proved. (3) the feature extraction algorithm based on local mean decomposition (LMD) and principal component analysis (PCA) denoising algorithm based on phase space reconstruction is studied. After the noise reduction based on the principal component analysis (PCA) method of phase space reconstruction, the PF1 with the highest energy is decomposed by LMD, and the envelope spectrum is analyzed. The simulation and rolling bearing fault experiments show that the method can effectively extract the fault characteristics of the signal. (4) A rolling bearing state recognition algorithm combining multi-scale permutation entropy and support vector machine is studied. By calculating the permutation entropy values of the four state signals of rolling bearing at each scale, choosing the appropriate scale to construct the eigenvector, selecting a certain number of feature vector samples and classifying them by using support vector machine classifier. The results show that the method has a high recognition rate for the normal, inner ring, outer ring and rolling body faults of the rolling bearing.
【学位授予单位】：武汉科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TH133.33;TH165.3

【参考文献】