Cyclostationary Approach for Instabilities Detection and Condition Monitoring of Centrifugal Compressor

Abstract

This study intends to explore the capabilities of the cyclostationary approach for instabilities detection and operating conditions monitoring of centrifugal compressors. Cyclostationary approach offers powerful signal analysis methods, applicable to different processes. It was proven useful for analysis of vibration, acoustic and pressure data for systems exhibiting periodicity. Cyclostationarity has been used for extracting subtle changes between cycles of the periodic signal which could be used for condition monitoring. Recent research focuses on employing this method for fault indication. Cyclostationary approach has not been extensively used in the field of turbomachinery, except for a few cases when it was proven to give a better insight into flow structure than standard signal processing techniques and allow for the detection of instabilities in flow systems. Thus, the cyclostationary approach may be suitable for instabilities detection and condition monitoring in centrifugal compressors. This paper exploits various techniques employing a cyclostationary framework for instabilities detection and operating conditions monitoring with the use of pressure signals from the low-speed centrifugal compressor. The most prospective cyclostationarity-based indicators are applied for the detection of instabilities. Due to a lack of second-order cyclostationarity, the study confines to the analysis of first-order cyclostationarity strongly exhibited in the compressor pressure signal. First-order cyclostationarity analysis provides an indication of instabilities and working conditions differentiation, but due to time-domain sampling, it is not fully robust and reliable. The highest potential is perceived in the cyclostationary approach use to extract changes between cycles. Different measures of change in variability could serve as a valuable indicator of instabilities.