摘要
研究了基于高阶谱(主要是三阶谱)和主元分析的故障特征提取方法。分析和实验表明,不同状态设备的三阶谱特征具有显著的差别,通过主元分析提取数字特征,为神经网络自动识别故障打下良好基础。通过往复泵排出阀不同状态———正常、弹簧失效、阀芯磨损特征的提取发现其特征存在较大差别。神经网络诊断结果表明所提出的方法具有良好的应用前景。
The extracting fault features approach based on high-order spectral for pump valves of reciprocating pump was investigated, especially third order spectral analysis. Experiment show that there exist distinct differences of high-order spectral among different states： normal, discharge valve＇s spring failure and discharge valve＇s core abrasion. Furthermore, comparing with the traditional approach , the approach is more efficient, adaptive, which indicates the potential of high-order spectral techniques in machinery fault diagnosis.
出处
《流体机械》
CSCD
北大核心
2007年第2期46-50,17,共6页
Fluid Machinery
关键词
往复泵
三阶谱
特征提取
故障诊断
reciprocating pump
three-order spectral
feature extraction
fault diagnosis
