ICA-R算法在旋转机械故障信号提取中的应用

Application of ICA-R algorithm in extraction of fault signals from rotating machinery

能否有效地提取故障信号的特征是对旋转机械进行故障诊断准确与否的关键所在。本文在介绍独立分量分析方法的基础上研究了基于参考信号的独立分量分析(ICA-R)算法,首先针对轴承与齿轮两类典型的旋转机械故障信号进行仿真,然后模拟出四路观测信号,同时根据先验知识建立参考信号,最后运用ICA-R算法对观测信号进行特征提取,并与仿真的故障信号进行对比分析。结果证明,ICA-R算法在旋转机械故障信号的特征提取中具有很强的优势,可实现弱故障信号的有效提取。

Whether the feature of fault signal can be extracted effectively is the key step in fault diagnosis of rotating machinery.Based on the study of Independent Component Analysis（ICA） method this paper studies a new algorithm of ICA with Reference（ICA-R）.Firstly,two typical kinds of rotating machinery fault signals including bearing and gear fault are simulated.Four observation signals are then constructed with the reference signal established according to a priori knowledge.Finally,the feature extraction is well done through applying ICA-R algorithm to observation signals.Results show that ICA-R algorithm in rotating machinery fault signal feature extraction has strong superiority and can be used to effectively extract weak fault signals.