Abstract: The vibration signals of faulty rotary machinery systems are generally multicomponent AM-FM signals. Aiming at the limitations of conventional demodulation methods for extracting the fault features of multicomponent vibration signals, a new approach for mechanical fault diagnosis based on iterated Hilbert transform (IHT) is proposed. The principle of IHT is introduced. It is seen from the analysis for an arbitrary two-component signals that there are limitations in the direct estimation of instantaneous frequencies with the phase signals obtained by IHT, so a smoothed instantaneous frequency estimation method based on difference operator and zero-phase digital low-pass filtering is proposed, and the simulation testing shows that this method is higher in accuracy and speed compared with the self-adaptive segmentation algorithm and Hilbert-Huang transform. The analysis results of the roller bearing vibration signal with an outer race fault and gearbox vibration signal with an broken teeth fault show that the fault features of mechanical fault signals can be extracted efficiently by using the proposed approach.

Key words: Gianfelici transform, Mechanical fault diagnosis, Multicomponent demodulation, Zero-phase filtering