Weak characteristic determination for blade crack of centrifugal compressors based on underdetermined blind source separation

An impeller is the core component of a centrifugal compressor. Therefore, the realization of detecting blade crack of centrifugal compressors in time is of great significance for industrial production. Different from bearings and gearboxes, the signal containing the effective information of blade vibration cannot be easily obtained due to the special complex working conditions of compressors. Considering the interaction between the blade and fluid structure, in this research, pressure pulsation (PP) signal of airflow near the rotating impeller inside the centrifugal compressor is investigated to determine the blade crack information. Then, an improved underdetermined blind source separation(UBSS) algorithm based on sparse component analysis(SCA) is proposed and applied on the sampled PP signal to obtain separated signals. Envelope analysis method is used for the separated signals next, and finally, the weak fault characteristic frequency of the cracked blade can be successfully determined with spectral analysis. The experimental result shows that the proposed UBSS-SCA algorithm together with PP signal can be used for the weak fault characteristic detection and long-term health monitoring of centrifugal compressor blades.