Fourier-Bessel series expansion based blind deconvolution method for bearing fault detection

In the last few years blind deconvolution techniques proved to be useful in order to extract impulsive patterns related to bearing fault from noisy vibration signals. Recently, a novel blind deconvolution method based on the generalized Rayleigh quotient has been proposed and an iterative algorithm related to the maximization of the cyclostationarity of the source has been defined. This paper presents a new condition indicator that exploit the Fourier-Bessel series expansion for the computation of a new cyclostationarity index that drives the maximization problem for the extraction of the excitation source. The main target of this work is to compare the results obtained through the exploitation of the Fourier-Bessel transform with respect to the classic Fourier transform in term of lower number of cyclic frequencies required for the algorithm. The comparison between the application of the two different methods involves both simulated and real signal taking into account a bearing fault. The results prove the capability of the new indicator to extract the impulsive source without the need of a set of cyclic frequencies but only with the first one, with a strong reduction of the computational time