Incipient fault diagnosis of bearings based on parameter-optimized VMD and envelope spectrum weighted kurtosis index with a new sensitivity assessment threshold

Due to difficulties in identifying localized and incipient bearing faults, most proposed fault diagnosis methods focus on detecting these faults. However, it is not clear to what extent of fault severity the proposed methods are capable of detecting. In other words, the crucial issue remains in the literature as to what is the criteria for defining an incipient defect for the proposed methods. This study attempts to address this challenge concerning a decomposed-based fault diagnosis method and provide a suitable measure for assessing this method. In this regard, a parameter-optimized VMD approach is used to decompose vibration signals. Proposed optimization algorithm is able to optimize VMD parameters so that the decomposed modes have the minimum bandwidth and noise interference. A new fault-sensitive index called the envelope spectrum weighted kurtosis index (WKI) is then implemented to detect the mode with the most fault information. This index has the highest sensitivity to fault symptoms and detects the most similarity between the original signal and decomposed modes. For introduced index, a related criterion called the sensitivity threshold (Sth) is given. Based on this criterion, the maximum effectiveness of the proposed method or the minimum observable fault severity can be addressed For validation, the proposed parameter-optimized VMD and the established index are challenged by the investigation of simulated vibration signals of a defective bearing at different fault severity and two experimental datasets and comparison with available methods in the literature. (C) 2020 ISA. Published by Elsevier Ltd. All rights reserved.