基于分段聚类的滚动轴承故障诊断方法研究

发布时间：2018-10-18 13:42

【摘要】：滚动轴承是旋转机械中极易损坏的元件之一。据统计,大约有30%的旋转机械故障是由于轴承故障引起的。因此,对滚动轴承进行状态监测与故障诊断具有很重要的意义。轴承故障冲击信号是持续时间很短的脉冲状波形,本文尝试将冲击信号从背景噪声中截取出来,通过计算冲击产生的频率,来判断冲击的来源,从而检测和诊断轴承工作状态。本文主要研究内容如下： (1)从滚动轴承振动信号的产生机理出发,讨论轴承运行时产生的各种振动成分。并对轴承工作时各种影响因素进行阐述。对轴承不同元件发生故障时产生的冲击脉冲特征进行研究。为滚动轴承故障诊断提供理论依据。 (2)对小波分析方法和傅立叶变换分析方法进行对比分析。应用小波分析实现对故障振动信号进行瞬态检测。在对冲击脉冲进行检测基础上对各个脉冲进行分段截取。 (3)对分段信号进行特征提取,包括时域特征、频域特征、小波包能量特征。由于这些特征是从不同角度反映了冲击脉冲的特性,但是,不同特征值反映脉冲性质的能力也是不一样的。由此需要引入主分量分析。通过介绍主分量分析算法原理、几何意义,在特征提取后应用主分量分析,结果使得少数几个主分量就能很好地反映不同冲击成分脉冲的性质,达到降维的目的。 (4)介绍聚类算法在轴承故障诊断中的应用。详细地介绍聚类算法中的距离定义、聚类准则函数的定义以及聚类算法的分类,比较不同聚类算法的特点。讨论模糊C均值聚类算法中模糊指数和聚类数两参数选取对聚类结果的影响。最后,通过两个模拟轴承故障实验来验证本文研究方法的有效性。一个是对外圈单一故障类型进行诊断,另一个是对外圈故障和滚动体故障的混合故障类型进行诊断。本文研究表明,基于分段聚类的轴承故障诊断方法是可行的,算法运算简单、可靠,对轴承进行精确的诊断具有很重要的意义。
[Abstract]:Rolling bearing is one of the most vulnerable components in rotating machinery. According to statistics, about 30% of rotating machinery failures are caused by bearing failures. Therefore, the condition monitoring and fault diagnosis of rolling bearings is of great significance. Bearing fault shock signal is a pulse waveform with short duration. This paper attempts to intercept the impact signal from the background noise and determine the source of the impact by calculating the frequency of the impact, so as to detect and diagnose the working state of the bearing. The main contents of this paper are as follows: (1) based on the mechanism of vibration signal of rolling bearing, the vibration components of bearing are discussed. And the bearing work of various factors are described. The impact pulse characteristics of different bearing components are studied. It provides theoretical basis for rolling bearing fault diagnosis. (2) the wavelet analysis method and Fourier transform analysis method are compared and analyzed. The transient detection of fault vibration signal is realized by wavelet analysis. On the basis of impulse detection, each pulse is segmented. (3) feature extraction of segmented signal, including time domain feature, frequency domain feature, wavelet packet energy feature. Because these characteristics reflect the characteristics of impulse pulses from different angles, however, the ability of different eigenvalues to reflect the properties of pulses is also different. It is necessary to introduce principal component analysis (PCA). By introducing the principle of principal component analysis (PCA) algorithm and geometric meaning, applying PCA after feature extraction, the results show that a few principal components can well reflect the properties of different impulse components. The purpose of dimension reduction is achieved. (4) the application of clustering algorithm in bearing fault diagnosis is introduced. The definition of distance, the definition of clustering criterion function and the classification of clustering algorithm are introduced in detail, and the characteristics of different clustering algorithms are compared. The influence of two parameters of fuzzy index and clustering number on the clustering results in fuzzy C-means clustering algorithm is discussed. Finally, the effectiveness of this method is verified by two simulated bearing failure experiments. One is to diagnose the single fault type of outer ring, the other is to diagnose the mixed fault type of outer ring fault and rolling body fault. This paper shows that the method of bearing fault diagnosis based on piecewise clustering is feasible, the algorithm is simple and reliable, and the accurate diagnosis of bearing is of great significance.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH133.31;TH165.3

【参考文献】