Quasi-Periodic Component Decomposition: New Decomposition Theory for Gear Fault Diagnosis

In this article, a new decomposition theory, quasi-periodic component decomposition (QPCD), is tailored for the feature extraction of machinery faults. The proposed QPCD defines the Blaschke transform, which employs a greedy iterative algorithm to represent complex signals as a linear combination of orthogonal Blaschke monocomponents with quasi-periodicity in Hardy space. Its derived Blaschke spectrum intuitively displays the energy distribution of Blaschke monocomponents, facilitating the analysis of signal intrinsic characteristics. Building upon this foundation, QPCD screens the useful monocomponents according to the dissimilarity between noise and quasi-periodic components within the Blaschke spectrum and recombines the useful Blaschke monocomponents based on period similarity, significantly attenuating noise interference. Through envelope spectrum analysis, it is observed that QPCD effectively mitigates noise interference, providing valuable assistance in the diagnosis of gear faults.