基于LMD的振动信号处理及故障特征提取研究

发布时间：2018-06-27 23:43

本文选题：小波阈值降噪 + LMD　；参考：《内蒙古大学》2015年硕士论文

【摘要】：旋转机械设备在现代化工业生产中发挥着不可小觑的作用,其正常运转是安全生产的重要保障。运用新兴的信号处理与故障特征提取方法对旋转机械进行实时检测与故障诊断,能在一定程度上保障机械设备系统安全高效的运行。滚动轴承是旋转机械设备的重要组成部分与典型代表,本文针对滚动轴承故障振动数据进行信号处理、故障特征提取与初步诊断的研究。实际采集到的振动信号中不仅包含轴承振动信号,而且还混入大量的噪声信号,这将严重影响信号分解以及后续的故障特征提取。因此,采用一定的方法进行信号降噪预处理则显得非常必要。本文以信噪比、最小均方根误差作为降噪性能的判别依据,通过仿真实验确定了最适合本文振动信号的小波函数、分解层数、阈值选择规则以及阈值函数并将其用于振动信号的降噪处理。鉴于LMD算法的自适应信号分解等特点,将其用于降噪振动信号分解得到乘积函数分量。通过分析故障振动信号的特点,提出了采用能量熵、奇异值熵、峭度以及Lemple-Ziv复杂度算法,从乘积函数分量中提取出故障特征并用于初步故障诊断。实验结果表明,轴承在正常运行状态下的能量熵和奇异值熵均大于内圈、滚动体以及外圈三种故障状态下的对应值,而峭度以及Lemple-Ziv复杂度指标在正常状态下要小于故障状态。综上所述,本文采用的小波阈值降噪方法能很好地实现信号的降噪预处理,运用LMD方法进行振动信号分解后提取的能量熵、奇异值熵、峭度以及Lemple-Ziv复杂度故障特征均能有效地实现滚动轴承的初步故障诊断。
[Abstract]:Rotating machinery plays an important role in modern industrial production, and its normal operation is an important guarantee of safe production. Using the new signal processing and fault feature extraction method to detect and diagnose the rotating machinery in real time can ensure the safe and efficient operation of the mechanical equipment system to some extent. Rolling bearing is an important part and typical representative of rotating machinery. In this paper, the signal processing, fault feature extraction and preliminary diagnosis of rolling bearing fault vibration data are studied. The actual vibration signals not only contain bearing vibration signals, but also mix with a large number of noise signals, which will seriously affect the signal decomposition and subsequent fault feature extraction. Therefore, it is very necessary to adopt certain method for signal denoising preprocessing. In this paper, the signal-to-noise ratio (SNR) and the minimum root mean square error (MMSE) are taken as the basis for judging the performance of noise reduction. The threshold selection rule and threshold function are applied to the noise reduction of vibration signal. In view of the characteristics of adaptive signal decomposition of LMD algorithm, it is used to decompose noise and vibration signal to get the product function component. By analyzing the characteristics of the fault vibration signal, an algorithm using energy entropy, singular value entropy, kurtosis and Lemple-Ziv complexity algorithm is proposed to extract the fault features from the product function component and apply them to the preliminary fault diagnosis. The experimental results show that the energy entropy and singular value entropy of the bearing under normal operation are larger than those of the inner ring, the rolling body and the outer ring, but the kurtosis and the Lemple-Ziv complexity index are smaller than the fault state in the normal state. To sum up, the wavelet threshold de-noising method used in this paper can achieve the signal de-noising preprocessing well, and the energy entropy, singular value entropy extracted from vibration signal decomposition by using LMD method, Kurtosis and Lemple-Ziv complexity fault features can effectively realize the initial fault diagnosis of rolling bearings.
【学位授予单位】：内蒙古大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TH165.3

【参考文献】