Research on Rolling Bearing Fault Diagnosis Method Based on Harmonic Noise Kurtosis-Time Characteristic Blind Deconvolution

Blind deconvolution filtering methods are a widely employed technique in the domain of mechanical fault diagnosis. However, they are frequently susceptible to interference signals, which have a considerable impact on their accuracy and stability. In light of the aforementioned limitations, this article introduces a novel blind deconvolution method, designated as harmonic noise kurtosis-time characteristic blind deconvolution (HTBD). This method comprises two principal components: The harmonic noise kurtosis (HNK) and time characteristic energy ratio (TCER) are two key components. HNK serves as an objective function in blind deconvolution, addressing both the joint periodicity and sparsity of the signal. The objective function is enhanced in terms of precision and stability through the incorporation of TCER, which facilitates the identification of fault periods and improves the effectiveness of the blind deconvolution process. Subsequently, we propose a blind deconvolution filtering method utilizing an iterative algorithm. Simulation and comparative experiments demonstrate that HTBD exhibits superior robustness against various interference signals and offers distinct advantages in terms of convergence speed and stability.