Maximum average kurtosis deconvolution and its application for the impulsive fault feature enhancement of rotating machinery

Blind deconvolution (BD) is a popular tool for vibration analysis, which has been extensively studied to extract useful information from contaminative signals for the diagnosis of rotating machinery. However, due to the disturbance of diverse interferences, good performance of conventional BD methods is usually hard to be guaranteed in some situations. Especially, when the rotating speed is time-varying, some advanced methods like maximum correlated kurtosis deconvolution (MCKD) and multipoint optimal minimum entropy deconvolution adjusted (MOMEDA) are even impracticable. To address these issues, the maximization of a new index named average kurtosis (AK) is treated as the objective function in this paper for deconvolution, i.e. maximum average kurtosis deconvolution (MAKD). AK inherently highlights the periodic impulses from angular domain, which is not only robust to some typical interferences, but also compatible with the variable speed condition. In this framework, an optimized Morlet wavelet is employed as the initial filter in the deconvolution process, which contributes to improving both the efficiency and performance of MAKD. The simulation analysis is conducted to demonstrate the robustness and capability of proposed method compared with several popular deconvolution methods, and experimental cases involving the failures of bearing and gear are further analyzed to clarify its practicability. (C) 2020 Elsevier Ltd. All rights reserved.

Automated summary

A new blind deconvolution method, MAKD, is proposed in this paper to address the issue of poor performance of conventional methods in certain situations. The method utilizes average kurtosis as the objective function, highlighting periodic impulse signals and showing robustness and compatibility with variable speed conditions. Simulation analysis and experimental cases are used to demonstrate its superior performance.