【作者】 孟涛；

【导师】 廖明夫；

【作者基本信息】 西北工业大学 ， 航空宇航推进理论与工程， 2003， 博士

【摘要】 振动分析是进行齿轮和滚动轴承的状态监测与故障诊断的重要手段。本文旨在研究适于齿轮和滚动轴承的振动信号处理方法。为此，分别建造了齿轮、滚动轴承故障诊断实验装置，模拟了9种齿轮故障、18种滚动轴承故障对所建立的故障诊断方法进行了验证和校核。 齿轮振动信号中常常含有大量噪声。为了有效地提取故障特征信息，本文建立了周期分段技术，即单齿分析技术。它将振动平均信号等距分段(所分段数与被监测齿轮齿数相等)，然后对分段信号进行频谱分析等。比较各段信号的异同以揭示故障齿的确切位置。实验分析结果表明，该方法对齿轮的表面局部损伤故障的早期诊断和断齿故障早期诊断十分有效，可以准确地辨别出有故障齿的位置，为齿轮故障诊断提供了一种有效手段。 齿轮和滚动轴承故障振动信号均呈现调制特征。因此，对测得的振动信号进行解调是齿轮和滚动轴承故障诊断的关键。常用包络解调技术对此类信号进行分析，但易受噪声影响，使得齿轮或滚动轴承的故障特征难以凸显。本文从研究相关函数的特性入手，经理论证明，发现自相关函数并不改变调制信号的调制特征，但具有显著的降噪特点。由此建立了时延相关解调方法，用于诊断齿轮和滚动轴承故障。这一方法的实现步骤是首先对测得的振动信号进行自相关分析，再对自相关函数进行时延，然后对时延后的自相关函数进行Hilbert变换解调。实验分析结果证实了时延相关解调技术是一种良好的降噪解调技术。并初步总结出了利用时延相关解调法识别滚动轴承不同故障类型的诊断特征。 另外，论述了常用的时频分析方法：Wigner-Ville分布与小波变换。在齿轮故障诊断中，Wigner-Ville分布不但可以判断局部故障的有无，且可监测局部故障的发展趋势。而在滚动轴承故障诊断中，Wigner-Ville分布难以判断故障的有无。Wigner-Ville分布计算时间长，不适于齿轮与滚动轴承的实时诊断。利用小波变换方法，分别对故障齿轮和滚动轴承的实测振动信号进行了分析、研究。结果表明：连续小波变换(CWT)可以诊断齿轮局部故障，但是其计算耗时，不宜用于实时诊断；离散小波变换(DWT)可以诊断滚动轴承故障，且可分辨轴承内环故障、外环故障及滚珠故障。但利用时延相关解调技术比使用DWT分析滚动轴承故障更为有效，且物理解释明确。 文中还针对转速、测点位置、故障分布(位置)、负载等因素对频谱分析、包络分析、时延相关解调等方法的诊断效果进行了大量的实验研究。结果表明，时延相关解调谱比包络谱受噪声影响小，能较好地凸显微弱故障信息。更多还原

【Abstract】 Detection of faults in gears and rolling element bearings has been paid much attention and deserves further investigation because of the diversity of the types and work conditions of these mechanical parts. Vibration analysis is widely used in the condition monitoring and fault detection of the gears and rolling element bearings. Many researches on this topic have been made[20-60], but much still remains to be done. For example, one of the problems is to suppress noise in the vibration signals measured on a gearbox or on a bearing pedestal. It is particularly important in demodulating the vibration signals. It is well known that the modulation of vibration signals is a decisive signature of the present of a local fault in either gears or bearings. The demodulation of these signals will more clearly reveal the signatures. However the result of demodulation is often masked by noise containing in the measured signals. Vibration signal processing techniques are sought in this dissertation. So the test rigs of the gear and the rolling element bearing are built. And different types of the faults are simulated individually in the gear and the rolling element bearing.In this paper, several conventional methods for processing the vibration signals are verified for detecting faults in gears and ball bearings with two test rigs designed especially for this investigation. The results of the experiment show that the peak value, RMS, standard deviation, crest factor, kurtosis value and impulse value of the vibration signals can distinguish between the normal and abnormal conditions of the ball bearing when the level of noise is low. However, why and how the faults happen in the machines can not be revealed. Then the resonance demodulation technique is used in analyzing the vibration signal of the gear and ball bearing. To some extent the results depend on the selected frequency band. In order to effectively extract the defect signature of the vibration signal, two new methods, single gear tooth analysis (period segmentation technique) and time-delayed correlation demodulation, are established and experimentally tested.The main idea of the single gear tooth analysis, is that the vibration signals collected with high sampling rate are divided into a number of segments with a same time interval. The number of the signal segments is equal to that of gear teeth. The analysis of individual segment reveals more sensitively the changes of the vibration signals in both time and frequency domain caused by gear faults. In addition, the location of failed tooth can be indicated in terms of the position of the segment that deviates from normal segments. An experimental investigation verified the advantages of the single gear tooth analysis. It is a good method for more sensitive detection of the incipient faults and locating the faults in the gear.The time-delayed correlation demodulation is established in order to suppress the noise and demodulating the signal. The auto-correlation functions of vibration signals measured on bearing cases are first computed, which will reduce the noise greatly, but not change the modulation signature of the signals. Then the auto-correlation functions are delayed for some time lags in order to decrease the affection of noise before demodulated by Hilbert Transform. The effectiveness of this method is confirmed by simulated data and experimental data. Moreover, thefaults on the outer ring, inner ring and rolling element can be recognized by the time-delayed correlation demodulation.Moreover, the time-frequency analysis methods, Wigner-Ville distribution and wavelet transform, are studied in this paper. Some faults in the gear and ball bearing can be detected by these two methods, however, the computing time is some long. So they can not be used in on-line fault detection and condition monitoring. And the explanation of the time-frequency analysis is not easily accepted by the engineers.The experimental factors, such as rotational speed and measured position and fault distribution a更多还原