数据驱动滚动轴承故障诊断研究

发布时间：2018-04-02 15:02

本文选题：故障诊断　切入点：滚动轴承　出处：《沈阳大学》2012年硕士论文

【摘要】：滚动轴承是现代工业生产中最为常用的一种部件,尤其是在旋转机械当中使用更为广泛。滚动轴承高发的故障率,给生产过程带来了巨大的影响。它威胁到生产安全,同时也会对经济利益造成损失。因此,对滚动轴承故障诊断的研究十分具有意义。本文主要以功率谱分析和数据驱动方法为理论基础,提出了利用主成分分析和费舍尔判别分析的方法来研究功率谱分析的滚动轴承的各类振动信号。主要研究工作如下: 1.本文对滚动轴承的故障机理及振动特征进行了详细分析,对几种主要的轴承故障形式以及主要的故障诊断方法进行了讨论分析。采用频域分析方法对轴承振动信号进行特征提取,并进一步利用主成分分析法对提取的特征进行筛选,降低信号特征向量的维数,以提高故障诊断的准确性和实时性。 2.滚动轴承故障发生时,其振动信号在某些频段内的信号能量分布会出现变化。分别根据正常数据和不同故障数据构建故障诊断模型,然后用T 2统计和Q统计的方法来检测不同的故障。通过对比识别准确率,对不同数据构建的模型的优劣进行了对比分析。 3.为了解决对不同轴承故障进行分类的问题,提出利用费舍尔判别分析在分类中最大化类间、最小化类内离散度的特点,可以对不同位置的故障以及相同位置不同大小的故障进行分类。在仿真试验中分类准确率比较理想。仿真实验结果表明,本文所提出的方法能较为准确地分辨出轴承的正常和故障状态,并对故障的位置及大小进行识别,可以较好地解决滚动轴承故障诊断的问题。
[Abstract]:Rolling bearing is one of the most commonly used parts in modern industrial production, especially widely used in rotating machinery.The high failure rate of rolling bearing has brought great influence to the production process.It is a threat to production safety, but also to the loss of economic benefits.Therefore, the study of rolling bearing fault diagnosis is of great significance.Based on the theory of power spectrum analysis and data-driven method, this paper presents a method of principal component analysis and Fisher discriminant analysis to study the vibration signals of rolling bearings based on power spectrum analysis.The main work of the study is as follows:1.In this paper, the fault mechanism and vibration characteristics of rolling bearing are analyzed in detail, and several main bearing fault forms and main fault diagnosis methods are discussed and analyzed.In order to improve the accuracy and real time of fault diagnosis, the feature extraction of bearing vibration signal is carried out by frequency domain analysis, and the feature extracted is screened by principal component analysis (PCA) to reduce the dimension of signal feature vector.2.When the rolling bearing fault occurs, the signal energy distribution of the vibration signal in some frequency bands will change.Fault diagnosis models are constructed according to normal data and different fault data, and different faults are detected by T 2 statistics and Q statistics.By comparing the accuracy of recognition, the merits and demerits of the models constructed by different data are compared and analyzed.3.In order to solve the problem of classifying different bearing faults, Fisher discriminant analysis is proposed to maximize inter-class and minimize intra-class dispersion.Faults in different locations and different sizes in the same position can be classified.The classification accuracy is ideal in the simulation experiment.The simulation results show that the method proposed in this paper can accurately distinguish the normal and fault state of the bearing, and identify the position and size of the fault, which can solve the problem of fault diagnosis of rolling bearing.
【学位授予单位】：沈阳大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH165.3

【参考文献】