基于分形分析的轴承故障状态分类研究

发布时间：2018-04-18 03:11

本文选题：状态监测 + 故障诊断　；参考：《中国科学技术大学》2011年硕士论文

【摘要】：在轴承的状态监测与故障诊断研究中,最主要的就是要找到最能体现轴承故障本质的特征量,故障特征的选则和提取是轴承故障研究中的关键之一。本论文选取能体现信号复杂程度的分形维数作为特征量,并分别从信号的时域和频域来计算信号的分形维数。第一章首先阐述了轴承故障诊断技术的选题背景与意义,以及轴承故障诊断系统中主要的研究内容;然后简要地介绍和比较了现有的检测方案与所采用的技术手段;然后分析了振动信号分析的处理方法。包括传统振动信号处理方法:时域统计量分析、Fourier分析、倒频谱分析、包络分析等方法;现代信号处理方法:短时Fourier变换、小波变换、双线性时频分析、循环平稳分析、分形分析,并分析了各种方法的优缺点与适应范围。第二章首先介绍了轴承的主要失效类型包括:磨损失效、疲劳失效、断裂失效、塑性变形失效等。接着重点计算了轴承在不受轴向力和受轴向力时的外圈故障特征频率、内圈故障特征频率和滚子故障特征频率。最后研究了轴承发生故障时,轴承振动信号的时域特性。第三章选用了一种与盒维数等价的网格维数来进行轴承振动信号的特征提取,通过改变采样点数,计算出选取不同采样点数的振动信号的网格维数,由不同的采样点数计算出的网格维数组成轴承状态的特征向量,并进一步建立轴承状态模式空间,根据未知轴承状态的特征向量与模式空间中的特征向量之间的距离大小来判断未知状态与已知状态的接近程度。第四章主要研究了基于小波变换的分形方法在轴承状态分类中的应用。首先对信号进行二进离散小波变换,再通过二进离散小波反变换得到不同尺度下的细节信号,根据原始信号的功率谱特点与1/ f过程的关系,选取1/ f过程的分形维数作为特征量,通过计算这些尺度下各细节信号的方差,再取对数,再进行直线拟合,得到斜率,根据斜率求出分形维数,通过对分形维数的分类再对轴承状态进行分类。根据最后的计算结果,此种方法可以有效的区分轴承的正常、外圈故障、滚珠故障这三种状态。第五章总结本论文的成果与不足并提出了研究展望。
[Abstract]:In the research of bearing condition monitoring and fault diagnosis, the most important thing is to find the characteristic quantity which can best reflect the nature of bearing fault. The selection and extraction of fault feature is one of the key points in bearing fault research.In this paper, the fractal dimension, which can reflect the complexity of the signal, is selected as the characteristic, and the fractal dimension of the signal is calculated from the time domain and the frequency domain respectively.The first chapter introduces the background and significance of bearing fault diagnosis technology, and the main research contents of bearing fault diagnosis system, and then briefly introduces and compares the existing detection schemes and technical means.Then the processing method of vibration signal analysis is analyzed.Traditional vibration signal processing methods include time-domain statistical analysis Fourier analysis, cepstrum analysis, envelope analysis, and modern signal processing methods: short time Fourier transform, wavelet transform, bilinear time-frequency analysis, cyclic stationary analysis, etc.Fractal analysis and analysis of the advantages and disadvantages of various methods and the scope of adaptation.The second chapter introduces the main failure types of bearing, such as wear failure, fatigue failure, fracture failure, plastic deformation failure and so on.Then, the fault characteristic frequency of outer ring, inner ring and roller are calculated.Finally, the time domain characteristic of bearing vibration signal is studied.In chapter 3, a grid dimension equivalent to the box dimension is selected to extract the feature of the bearing vibration signal. By changing the sampling points, the grid dimension of the vibration signal with different sampling points is calculated.The characteristic vector of bearing state is made up of the grid dimension calculated by different sampling points, and the bearing state mode space is further established.According to the distance between the eigenvector of the unknown bearing state and the eigenvector in the pattern space, the degree of proximity between the unknown state and the known state is determined.In chapter 4, the application of fractal method based on wavelet transform in bearing state classification is studied.Firstly, the signal is transformed by dyadic discrete wavelet transform, and then the detail signals of different scales are obtained by using the dyadic discrete wavelet inverse transform. According to the power spectrum characteristics of the original signal and the relationship between the 1 / f process and the power spectrum of the original signal,The fractal dimension of 1 / f process is selected as the characteristic quantity. By calculating the variance of each detail signal under these scales, then taking the logarithm, the slope is obtained by straight line fitting, and the fractal dimension is calculated according to the slope.The state of bearing is classified by the classification of fractal dimension.According to the final calculation results, this method can effectively distinguish the normal bearing, the outer ring fault and the ball fault.The fifth chapter summarizes the achievements and shortcomings of this paper and puts forward the research prospect.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3

【引证文献】