The relationship between fault-induced impulses and harmonic-cluster with applications to rotating machinery fault diagnosis

摘要

The vibration signal measured from rotating machinery occurring faults will contain the fault-induced impulses with frequency modulation effects, from which we can extract the corresponding fault features. Besides, we can also observe a specific spectral distribution equipped with the fault-featured frequency interval within the resonance band of fault-related vibration signal. To explore the underlying relationship between these two kinds of fault-featured patterns, this paper bridges the equivalence between fault-induced impulses and a harmonic-cluster (i.e., a group of harmonics within the specific frequency band) with uniform interval in frequency-domain for the first time. Such an established relationship not only reveals the consistency of fault features implied in modulation patterns and spectral distribution of fault-related vibration signal, but also provides a harmonic-decomposition based strategy to detect the fault-induced cyclic impulses within identified resonance band. Compared with the results obtained by other advanced approaches, the proposed adaptive harmonic-decomposition method is able to detect the fault-induced impulses with higher signal-to-noise ratio (SNR) in time-domain and more distinct fault-featured structure in envelope spectrum, regardless of low-SNR simulated cases and two fault-related real vibration signals.