谱相关特性在机械信号特征提取中的应用研究

将机械振动信号视为循环平稳信号，用循环统计方法进行处理。探讨了循环平稳信号的遍历性及其统计量的时间平均估计方法，着重研究了二阶循环统计量的谱相关特性，指出利用谱相关特性的频率选择性和噪声抑制性可以消除信号中的加性或乘性随机噪声以及非同频成分的干扰，有效地提取出信号特征频率。通过对滚动轴承、注浆泵等旋转和往复机械的振动信号分析表明，谱相关特性在机械信号特征提取方面具有很好的应用价值。

Spectral Correlation in Mechanical Signal Feature Extraction

Considering the signals as cyclostationary signals, the paper discussed the cycloergodic ity of cyclostationary signal and the time--average estimation of cyclic statistics. The research focused on the spectral correlation characteristics of second--order cyclic statistics, which were the frequency selectivity and the noise elimination, Using the characteristics to process the signals, the additive or the multiple random noise and the periodical interference in the signals can be eliminated, so the fea- ture frequency are extracted effectively, These advantages are proven by the real mechanical vibration signals from the rolling bearing and the injection pump. The results show that the spectral correlation characteristics are effective in the feature extraction of mechanical signals.