Signal analysis for rotating machinery vibrations

Abstract A novel method is proposed for the purpose of characterizing repetitive vibration signals. This consists in decomposing the signal into periodic—not necessarily harmonic—components. Two parameters are then sufficient for the description of each component. The appropriate signal processing technique is based on time domain averaging. The proposed analysis is shown to be valuable for cases where the usual techniques of spectrum analysis fail to indicate a vibratory signature. Applications indicated include hypothetical bearing and gear vibrations.