基于非平稳时序分析的滚动轴承故障诊断方法研究

发布时间：2018-10-22 11:28

【摘要】：滚动轴承振动信号是由不同零部件振动耦合而成，且滚动轴承工作环境极为复杂，因此具有非常明显的非平稳性，如何从振动信号中提取能够准确反应滚动轴承工作状态的特征量成为对其进行故障诊断关键所在。本文以对滚动轴承振动信号的分析为基础，重点研究了基于时序分析的经验模态分解和时变自回归模型方法在滚动轴承故障诊断中的应用。针对经验模态分解后选择固有模态函数时具有盲目性的问题，提出一种基于能量阀值的固有模态函数筛选方法，实现了对各固有模态函数进行自适应筛选的目的，进而实现对信号故障信息的重构，为进一步的特征提取打下基础。针对传统的EMD能量特征提取方法在滚动轴承故障诊断中识别率较低的问题，通过将正常信号视为历史信号，提出一种提取相同工况条件下正常信号与故障信号各固有模态函数能量比构建特征向量的特征提取方法，对有效提取滚动轴承的运行特征，进而实现轴承的在线故障诊断，提供了一种新的解决思路。针对传统TVAR模型特征提取中存在建模不准确的问题，，提出一种提取建模过程中的模型阶次判定值构建特征向量的特征提取方法，从而避免了对不同样本信号建立相同阶次模型所造成模型不准确的问题，对有效提取滚动轴承的运行特征提供了新的解决思路。仿真实验表明，本文所提出的非平稳信号特征提取方法能够有效地、准确地识别滚动轴承在不同工况条件下的不同故障类型。
[Abstract]:The vibration signal of rolling bearing is formed by the vibration coupling of different parts, and the working environment of rolling bearing is very complex, so it has very obvious non-stationarity. How to extract features from vibration signals that can accurately reflect the working state of rolling bearings is the key to fault diagnosis. Based on the analysis of the vibration signals of rolling bearings, this paper focuses on the application of empirical mode decomposition and time-varying autoregressive models based on time series analysis to the fault diagnosis of rolling bearings. In order to solve the problem of blindness in selecting inherent mode function after empirical mode decomposition, a method of selecting inherent mode function based on energy threshold is proposed, which realizes the purpose of adaptive screening of each inherent mode function. Furthermore, the reconstruction of signal fault information is realized, which lays the foundation for further feature extraction. In order to solve the problem of low recognition rate of traditional EMD energy feature extraction method in rolling bearing fault diagnosis, the normal signal is regarded as a historical signal. In this paper, a feature extraction method is proposed to extract the energy ratio of normal signal and fault signal to construct the eigenvector under the same working condition, which can effectively extract the running feature of rolling bearing and realize the on-line fault diagnosis of the bearing. A new solution is provided. Aiming at the problem of inaccurate modeling in traditional TVAR model feature extraction, a feature extraction method is proposed to construct the feature vector of model order decision value in the process of model extraction and modeling. Thus, the inaccuracy of the model caused by the establishment of the same order model for different sample signals is avoided, and a new way of solving the problem of effectively extracting the running characteristics of rolling bearings is provided. The simulation results show that the proposed non-stationary signal feature extraction method can effectively and accurately identify the different fault types of rolling bearings under different working conditions.
【学位授予单位】：湖南工业大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH165.3

【相似文献】