Development of Non-stationary Condition Indexes for Condition Monitoring of Rotating Machinery using Vibration Signals and its Application to a Large Centrifugal Compressor

Statistical measures, such as standard deviation(or RMS), skewness, kurtosis, and spectral power of a measured vibration signal, are often used as key features when implementing a condition-based maintenance(CBM) system, particularly for rotating machinery. Typically, the CBM system monitors these statistical values and compares them with the values from an initial healthy state. However, it is often the case that the initial reference vibration signals are not available, making the implementation of the CBM system more challenging. In this study, as an alternative method to overcome this situation, non-stationary condition indexes are developed by capturing the non-stationary characteristics of measured vibration signals. The suggested non-stationary condition indexes are based on time-varying statistical measures and are applied to the same centrifugal compressor as in the previously published paper. The results are consistent with the previous findings, showing that the proposed non-stationary condition indexes may be useful alternative criteria, especially when the initial reference vibration signals are not available.