基于VMD的滚动轴承故障诊断研究

发布时间：2018-03-30 12:07

本文选题：故障诊断　切入点：滚动轴承　出处：《兰州交通大学》2017年硕士论文

【摘要】：随着中国工业化进程不断推进,不断有生产机器开始进入老化期,在将来会达到一个庞大的数量。滚动轴承是旋转机械重要零部件之一,也是占比最大的故障源之一。因此,开展滚动轴承故障诊断研究具有重要的现实意义和经济意义。模态提取是滚动轴承故障诊断的关键,尤其是对滚动轴承故障特征的提取。滚动轴承振动信号属于典型的非线性信号,特征提取的质量直接影响故障诊断结果。针对故障特征提取与识别问题,研究内容如下:(1)通过介绍变分模态分解方法(Variational Mode Decomposition,VMD)中的本征模态函数、维纳滤波和解析信号的基本概念,叙述了如何构造变分模态分解方法中的信号约束问题,并随后介绍了如何使用变分模态分解方法如何求解约束问题。为了验证变分模态分解方法的优越性,分别用变分模态分解方法和经验模态分解方法对噪声干扰信号和脉冲干扰信号进行分解。结果表明,变分模态方法在噪声鲁棒性和脉冲干扰性上具有明显优势。(2)使用基于峭度准则VMD及平稳小波的轴承故障诊断方法,提取强噪声背景下的滚动轴承故障特征。首先使用变分模态分解对同一负荷下的故障信号进行预处理,再通过峭度准则筛选出最佳和次佳信号分量进行重构并使用平稳小波进行去噪处理,最后分析信号的包络谱来对轴承的故障类型进行判断。通过对仿真滚动轴承内圈故障信号进行分析,该方法可成功提取出微弱特征频率信息,噪声抑制效果优于EMD(Empirical Mode Decomposition,EMD)。由此表明,基于峭度准则VMD及平稳小波的轴承故障诊断可有效提取强声背景下的滚动轴承早期故障信息,具有一定的可靠性和应用价值。(3)使用基于VMD瞬时能量法及MPSO-SVM的轴承故障诊断方法,实现轴承振动故障的较精确诊断。首先使用变分模态分解方法分解轴承振动信号,再根据VMD分量特性筛选出包含主要故障信息的分量进行瞬时能量特性计算并构建故障特征向量,最后将其输入变异粒子群算法(Mutation Particle Swarm Optimization,MPSO)优化后的支持向量机(Support Vector Machine,SVM)分类器中来区分滚动轴承的工作状态和故障类型。对轴承正常状态、内圈故障及外圈故障信号进行仿真实验,该方法可较精确的对轴承振动信号进行故障分类,具有良好的分类效果。
[Abstract]:As China's industrialization continues to advance, more and more production machines begin to enter the aging period, which will reach a large number in the future. Rolling bearings are one of the important parts of rotating machinery and one of the biggest fault sources. The research of rolling bearing fault diagnosis has important practical and economic significance. Modal extraction is the key of rolling bearing fault diagnosis. The vibration signal of rolling bearing is a typical nonlinear signal, and the quality of feature extraction directly affects the fault diagnosis result. The research contents are as follows: (1) by introducing the intrinsic mode functions, Wiener filtering and the basic concepts of analytical signals in the variational Mode decomposition method (VMD), the paper describes how to construct the signal constraint problem in the variational mode decomposition method. Then it introduces how to use variational mode decomposition method to solve constraint problem, in order to verify the superiority of variational mode decomposition method. The variational mode decomposition method and the empirical mode decomposition method are used to decompose the noise interference signal and the pulse interference signal respectively. Variational mode method has obvious advantages in noise robustness and impulse interference. (2) the bearing fault diagnosis method based on kurtosis criterion VMD and stationary wavelet is used. The fault characteristics of rolling bearing under strong noise background are extracted. Firstly, the fault signals under the same load are preprocessed by variational mode decomposition. Then the best and sub-optimal signal components are selected by kurtosis criterion for reconstruction, and the stationary wavelet is used to Denoise the signal. Finally, the envelope spectrum of the signal is analyzed to judge the fault type of the bearing. By analyzing the fault signal of the inner ring of the rolling bearing, the weak characteristic frequency information can be extracted successfully by this method. The noise suppression effect is better than that of EMD(Empirical Mode Decomposition.Therefore, it is shown that bearing fault diagnosis based on kurtosis criterion VMD and stationary wavelet can effectively extract the early fault information of rolling bearing under strong sound background. The method of bearing fault diagnosis based on VMD instantaneous energy method and MPSO-SVM is used to realize the accurate diagnosis of bearing vibration fault. First, the variational mode decomposition method is used to decompose the bearing vibration signal. Then according to the characteristics of VMD components, the components containing the main fault information are selected to calculate the instantaneous energy characteristics and the fault feature vectors are constructed. Finally, the support vector machine (SVM) support Vector Machine (SVM) classifier is used to distinguish the working state and fault type of rolling bearing by input mutation Particle Swarm optimization (MPSO). The simulation experiments are carried out on the normal state of bearing, inner ring fault and outer ring fault signal. This method can classify bearing vibration signals accurately and has good classification effect.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TH133.33

【参考文献】