基于小波包和包络谱齿轮箱异响分析.docVIP

基于小波包和包络谱齿轮箱异响分析 　　摘 要 ：针对齿轮箱异响信号呈非平稳时变特征并伴随有强烈的背景噪声，介绍了基于小波包分解、频带能量分析和包络谱相结合的齿轮箱异响分析方法。首先对采集到的齿轮箱声学信号进行小波包分解，对该信号进行小波包能量化分，然后对照正常和异响发动机信号的能量特征向量，对明显差异的小波包系数进行重构，最后对重构信号进行包络分析提取故障特征频率为16.5#8197;Hz，与实际的故障特征频率相近，表明该方法适用于齿轮箱的故障分析。 　　关键词：小波包分解；包络分析；异响；频带能量；特征提取 　　中图分类号：TP206+.3文献标文献标志码：A文献标DOI：10.3969/j.issn.2095-1469.2011.05.008 　　 　　Analysis of Abnormal Sound in Gearbox Based on Wavelet Packet and Envelope Spectrum 　　Yang Cheng1，Xia Luning1，Yang Zhendong2 　　(1.School of Mechanical Engineering，Chongqing University，Chongqing 400044，China； 　　2.South Motorcycle Technology Research Institute，Chongqing 401320，China) 　　 　　Abstract：The abnormal sound signals of gearbox are non-stationary, time-varying and are accompanied by strong background noise. This paper introduced an analysis method of abnormal sound that combined the wavelet packet decomposition, the energy band analysis and envelope spectrum. The paper decomposed the acoustic signals from the gearbox with wavelet packet, quantified the signals by wavelet packet energy, compared the energy eigenvectors of normal sound with those of abnormal sound of the engine signals, and then reconstructed the significantly different wavelet packet coefficients. Finally, envelope analysis was applied to the reconstructed signals. The fault characteristic frequency, 16.5 Hz, which is similar to the actual fault frequency, was extracted. It is demonstrated this method was suitable for failure analysis. 　　Keywords：wavelet packet；envelope analysis；abnormal sound；band energy；feature extraction 　　 　　由于受到齿轮箱内部结构复杂，发动机转速不稳，壳体噪声辐射，线上诊断的强背景噪声等影响，使采集到的信号频谱非常复杂，是典型的非平稳信号[1]。这就对齿轮箱故障诊断的准确率提出了更高的要求，传统的信号处理技术已满足不了对微弱的故障特征提取、早期故障诊断及非平稳信号特征提取的要求。傅里叶分析是针对平稳信号的整个频段进行分析，对整体频谱的体现比较有效，但多次平均容易淹没微弱的特征，所以对非平稳复杂信号的特征提取显得捉襟见肘。随着计算机存储容量和运算速度的迅速提高以及信号分析理论和方法的研究开发，已有很多先进的信号分析方法在机械故障诊断中得到应用，其中小波包分解等先进的信号处理技术为故障诊断中这些难题的解决提供了理论基础。它为信号的时―频分析提供了有力的手段，可对信号的任意局部进行分析，而且其分析精度可变，是一种用加时窗进行分析的方法。在高频段，具有高的时间分辨率和低的频率分辨率，而在低频段，具有低的时间分辨率和高的频率分辨率[2]，克服了傅立叶变换中时频分辨率恒定的弱点。因此它享有“数字显