滚动轴承振动信号特征提取及诊断方法研究

发布时间：2018-05-26 20:03

本文选题：滚动轴承 + 故障诊断　；参考：《大连理工大学》2013年博士论文

【摘要】：滚动轴承是旋转机械中最常用的零部件之一,其工作状态正常与否直接影响到整台设备的性能甚至整个生产线的安全和生产。因此,研究滚动轴承的状态监测和故障诊断技术,对于避免突发事故的发生以及维修体制的变革等具有重要的理论价值和现实意义。本文以滚动轴承为研究对象,针对特征提取这一滚动轴承故障诊断中的关键问题,从轴承振动信号的处理着手,进行了一系列的研究工作。主要内容如下： 1)阐述了论文选题的背景与意义,分析了滚动轴承的振动机理,总结了滚动轴承诊断信息获取、故障特征提取和故障模式识别等方面国内外的研究现状,在此基础之上,确立了本文的研究思路和研究内容。 2)提出了一种基于IMF峭度的滚动轴承故障诊断方法。在分析峭度在轴承故障诊断中的局限性的基础上,提出利用经验模式分解(EMD)方法把滚动轴承的振动信号分解为一系列代表不同频带的‘内蕴模式函数(IMF),再取包含丰富故障信息的前几个IMF计算其峭度值,用不同频带信号的峭度值来进行轴承工作状态的分析,最后以这些峭度值作为故障特征向量,输入支持向量机(SVM)实现轴承的故障诊断,并用实验信号进行了验证。 3)针对IMF峭度对轴承故障损伤程度识别能力的不足,提出了一种基于IMF包络样本熵的滚动轴承故障诊断方法。首先介绍了熵概念的发展及泛化,并给出了样本熵的定义,然后针对样本熵在表征信号复杂度时存在熵值大小和复杂度高低不一致的缺点,结合滚动轴承故障信号的调制特征,提出利用包络信号的样本熵作为滚动轴承的故障特征。该方法利用EMD先把振动信号分解为若干IMF之和,再选取包含丰富故障信息的IMF求其包络信号,最后计算包络信号的样本熵,然后以包络样本熵组成故障特征向量,结合SVM完成滚动轴承不同工作状态的识别。结果表明,所提方法不仅能够区分轴承不同类型的故障,还能准确识别不同的故障损伤程度。 4)将层次熵(Hierarchical entropy,HE)引入到滚动轴承故障特征提取中。在分析样本熵和多尺度熵表征信号复杂度的不足的基础上,介绍了层次熵的基本概念和算法,提出了基于层次熵的滚动轴承故障特征提取方法。层次熵通过构造特定的算子,不但能够考察信号“低频成分”的样本熵,还能够计算“高频成分”的样本熵。最后,结合SVM实现滚动轴承的故障诊断,同时与前面提出的故障诊断方法以及多尺度熵方法进行了对比分析。实验验证结果表明,基于层次熵和SVM的滚动轴承故障诊断方法效果最优,识别率达到了100%。 5)提出了一种基于EMD和相关系数的滚动轴承早期故障检测方法。运用模式识别方法进行不同轴承故障类型的分类是建立在故障样本数据可获得的基础之上的,而在实际的生产应用中,很难获得不同故障类型的轴承振动数据。针对这一问题,把轴承故障检测当作是异常检测问题,完全基于轴承正常振动信号,通过相关系数的计算和简单的设定,实现轴承故障的自动检测。完成故障的检测以后,利用包络分析法完成轴承故障的诊断,最后通过全寿命周期实验信号验证了该方法的有效性。 6)将希尔伯特振动分解(Hilbert Vibration Decomposition,HVD)引入到滚动轴承的故障诊断中。在分析EMD模态混叠现象的基础上,介绍了HVD的基本原理和性质,并用仿真信号对比分析了HVD和EMD在分解含有异常事件的信号时的分解效果。然后,提出了基于HVD的滚动轴承故障诊断方法。该方法首先运用HVD对轴承振动信号进行分解,再结合包络分析实现轴承的故障诊断,实验信号的分析结果表明该方法能够有效地进行滚动轴承的故障诊断。
[Abstract]:Rolling bearing is one of the most commonly used parts in rotating machinery. Its normal working condition directly affects the performance of the whole equipment and even the safety and production of the whole production line. Therefore, it is important to study the state monitoring and fault diagnosis technology of the rolling bearings to avoid the occurrence of sudden accidents and the reform of the maintenance system. The paper takes the rolling bearing as the research object, and aims at the key problems in the fault diagnosis of the rolling bearing, and carries out a series of research work from the processing of the bearing vibration signal. The main contents are as follows:
1) this paper expounds the background and significance of the topic selection, analyzes the vibration mechanism of the rolling bearing, summarizes the research status of the diagnosis information acquisition, fault feature extraction and fault pattern recognition of rolling bearings, and based on this, establishes the research ideas and research contents of this paper.
2) a fault diagnosis method for rolling bearings based on IMF kurtosis is proposed. On the basis of analyzing the limitation of kurtosis in bearing fault diagnosis, the method of empirical mode decomposition (EMD) is proposed to decompose the vibration signals of rolling bearings into a series of 'intrinsic mode functions (IMF) "representing different frequency bands, and then contains rich fault information. The first few IMF calculated its kurtosis value and analyzed the working state of bearing with the kurtosis value of different frequency band signals. Finally, these kurtosis values were used as fault feature vectors, and the support vector machine (SVM) was input to realize the fault diagnosis of bearing, and the experimental signal was used to verify it.
3) in view of the lack of IMF kurtosis to identify the bearing fault damage degree, a fault diagnosis method based on IMF envelope sample entropy is proposed. Firstly, the development and generalization of entropy concept are introduced, and the definition of sample entropy is given. Then, the entropy value and complexity of the sample entropy are used to characterize the number and complexity of the letter number complexity. In this method, the sample entropy of the envelope signal is used as the fault feature of the rolling bearing. This method uses the EMD to decompose the vibration signal into the sum of several IMF, and then selects the IMF containing the rich fault information to obtain the envelope signal, and finally calculates the sample entropy of the envelope signal. Then, the sample entropy of the envelope signal is calculated, and then the sample entropy of the envelope signal is calculated. Then, the sample entropy of the envelope signal is calculated, and then the sample entropy is calculated. Then, the sample entropy of the envelope signal is calculated, then the sample entropy is calculated. Then, the sample entropy of the envelope signal is calculated. The fault feature vectors are composed of envelope sample entropy and SVM is used to identify different working states of rolling bearings. The results show that the proposed method can not only distinguish different types of fault of bearing, but also identify different degree of fault damage accurately.
4) Hierarchical entropy (HE) is introduced into the fault feature extraction of rolling bearings. Based on the analysis of the shortage of sample entropy and multi-scale entropy, the basic concepts and algorithms of hierarchical entropy are introduced, and the method of extracting the fault characteristics of rolling bearings based on hierarchical entropy is proposed. It can not only examine the sample entropy of the signal "low frequency components", but also calculate the sample entropy of the "high frequency component". Finally, the fault diagnosis of rolling bearings is realized with SVM. At the same time, it is compared with the previous fault diagnosis method and the multi-scale entropy method. The verification results show that the rolling based on the level entropy and the SVM are rolling. The bearing fault diagnosis method has the best effect and the recognition rate reaches 100%.
5) an early fault detection method of rolling bearing based on EMD and correlation coefficient is proposed. The classification of different bearing fault types using pattern recognition method is based on the data obtained from the fault sample data. In actual production application, it is difficult to obtain the bearing vibration data with different types of fault. The problem of bearing fault detection is regarded as an anomaly detection problem. It is based on the normal vibration signal of the bearing. The bearing fault detection is realized by the calculation of the correlation coefficient and the simple setting. After the fault detection is completed, the diagnosis of the bearing fault is completed by the envelope analysis method. Finally, the whole life cycle experimental signal is verified. The effectiveness of the method.
6) the Hilbert vibration decomposition (Hilbert Vibration Decomposition, HVD) is introduced into the fault diagnosis of rolling bearings. Based on the analysis of EMD modal aliasing, the basic principles and properties of HVD are introduced, and the decomposition effects of HVD and EMD in the decomposition of signals containing abnormal events are compared and analyzed by the simulation signals. The method of fault diagnosis of rolling bearing based on HVD. This method first decomposes the vibration signal of bearing by using HVD, and then combines envelope analysis to realize the fault diagnosis of bearing. The analysis result of experimental signal shows that the method can effectively diagnose the fault of rolling bearing.
【学位授予单位】：大连理工大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：TH133.33;TH165.3

【参考文献】