An Adaptive Harmonic Product Spectrum for Rotating Machinery Fault Diagnosis

Although the frequency band segmentation rules used in fast kurtogram (FK) has made great achievements in locating fault resonance band, it still has the problem of underdecomposition or overdecomposition for resonance band. This will affect the performance of the methods based on this band segmentation rule in fault diagnosis. Aiming at this problem, an adaptive harmonic product spectrum (AHPS) is proposed in this article, which can realize adaptive frequency band segmentation adaptively based on signal power spectral density (PSD) instead of fixed segmentation rules. AHPS can achieve adaptive frequency spectral segmentation by iteratively convolving the PSD with the Gaussian kernel function. The harmonic saliency index (HSI) is used to pave a spectrum plane. The optimal resonance band is determined by maximizing HSI and more accurate fault resonance band location performance is achieved. Both simulation data and measured data are used to verify the performance of the proposed AHPS. The results show that compared with FK and harmonic product spectrum (HPS), AHPS has better robustness and fault diagnosis performance.