Adaptive correlated Kurtogram and its applications in wheelset-bearing system fault diagnosis. (1st June 2021)

• Record Type: Journal Article
• Authors: Liu, Zechao; Yang, Shaopu; Liu, Yongqiang; Lin, Jianhui; Gu, Xiaohui
• Is Part Of: Mechanical systems and signal processing. Volume 154(2021)
• Publication Date: 2021-06-01
• Subjects: Wheelset-bearing system -- Scale-space representation -- Kurtogram -- Correlated kurtosis -- Empirical wavelet transform; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2020.107511
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: A novel method for arbitrary generating the paving of the plane of the Kurtogram is proposed. Correlated kurtosis is used for characterization the impulsive components of different periodic. Wheelset tread damage and wheelset bearing fault detection are achieved, simultaneously. Abstract: As one of the crucial subsystems, the performance of the wheelset-bearing system will significantly affect the running safety, stability and comfort of a high-speed train. Compared with the traditional rotating machinery vibration signals, those of wheelset-bearing systems have a distinct difference, for instance, the wheelset tread damage induced vibrations and bearing fault-related vibrations coexist and have similar wave-forms. Therefore, it is necessary to advance a reasonable method to identify the two kinds of cyclo -stationary components. To achieve this purpose, the adaptive correlated Kurtogram (ACK) is proposed. The proposed ACK method has two advantages: the first one is it can adaptively generate the paving of the plane for Kurtogram by using scale-space representation (SSR) to guarantee the detected boundaries cover the resonant frequency band well; the second one is it can highlight the special periodic impulses by correlated kurtosis. The scale-scale plane (SSP), which is obtained by SSR, is used to represent the frequency distribution characteristic of the acquired signal, firstly. After that, the paving of the plane is arbitrarily divided based on the SSP. Then the empirical wavelet transform (EWT) is employed to construct the sub-band signals according to the paving of the plane. The correlated kurtosis values with interesting periodic are calculated to distinguish the special repetitive transients. The nodes with the largest correlated kurtosis values are used to conduct a further envelope spectrum analysis. ACK method improves the effectiveness of the identification of the frequency bands containing impact information, and is capable of detecting multiple frequency bands simultaneously. The proposed ACK method is demonstrated by the simulated signal and two experimental signals acquired from a wheelset-bearing system test bench. The results indicate that the ACK can simultaneously extract the wheelset tread damage and bearing fault-related vibration component, and it is superior to the other traditional methods.