利用稀疏盲源分离方法的叶片裂纹特征提取

Blade crack feature extraction by using sparse blind source separation algorithm

叶轮是离心压缩机的核心部件,实现压缩机叶片裂纹早期故障识别在工业生产中具有非常重要的意义。有裂纹的叶片的异常振动会直接反映到流体的压力脉动中。然而实际中由叶片裂纹造成的异常振动非常小,使得压力脉动中的故障信息非常微弱,导致故障频率难以识别。先利用稀疏盲源分离方法对离心压缩机扩压器处的压力脉动信号进行处理,然后对分离的信号进行包络分析,最后提取出故障特征频率,实现了离心压缩机叶轮叶片裂纹故障检测,可以对叶轮状态进行长期实时监测。

The impeller is the core component of a centrifugal compressor, and so the realization of the identification of the compressor blade crack fault if of great significance in the industrial production. The abnormal vibration of the blade with cracks will be directly reflected in the fluid pressure pulsation. In practice, however, the abnormal vibration caused by blade cracks are very small, which usually means that it is very difficult to identify the fail frequency since the fault information in the pressure pulsation is too weak. In this paper we firstly deal with the fluid pressure pulsation signals which flow through the fault of impeller bleaching with the sparse underdetermined blind source separation algorithm. Then, the envelope analysis method is applied to the separated signals. Finally, we extract the fault characteristic frequency from the separated signals by the envelope analysis and successfully detect the blade crack fault on the centrifugal compressor impeller. This can be used for long- term health monitoring of the centrifugal compressor impeller.