Envelope Spectrum L-Kurtosis and Its Application for Fault Detection of Rolling Element Bearings
Wenjie Bao, Xiaotong Tu, Yue Hu, Fucai Li
Abstract
Wenjie Bao, Xiaotong Tu, Yue Hu, Fucai Li
Abstract
As one of the most important and essential elements of machines, rolling element bearings always fail due to the severe operating environment. Bearing failures usually result in the periodic impulses, which are the most crucial feature for the bearing diagnosis. These impulses may be overwhelmed by the background noises or other unrelated components. Many traditional features in the time domain such as kurtosis and root mean square (rms) are invalid in some cases. They are ineffective in detecting periodic impulses. This paper proposed a novel frequency-domain index, named envelope spectrum (ES) L-kurtosis (ESLK), to detect the periodicity of impulses. In this paper, simulations and bearing degradation data are utilized to clarify the properties of ESLK. In addition, this paper also proposed an ESLK-based method to locate the frequency range of periodic impulses. This method is verified by simulations and an experiment. The results illustrate that the proposed method performs better than the traditional kurtosis-based method.
OpenAlex reports 50 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
As one of the most important and essential elements of machines, rolling element bearings always fail due to the severe operating environment. Bearing failures usually result in the periodic impulses, which are the most crucial feature for the bearing diagnosis. These impulses may be overwhelmed by the background noises or other unrelated components. Many traditional features in the time domain such as kurtosis and root mean square (rms) are invalid in some cases. They are ineffective in detecting periodic impulses. This paper proposed a novel frequency-domain index, named envelope spectrum (ES) L-kurtosis (ESLK), to detect the periodicity of impulses. In this paper, simulations and bearing degradation data are utilized to clarify the properties of ESLK. In addition, this paper also proposed an ESLK-based method to locate the frequency range of periodic impulses. This method is verified by simulations and an experiment. The results illustrate that the proposed method performs better than the traditional kurtosis-based method.
Key concepts: Kurtosis, Rolling-element bearing, Envelope (radar), Bearing (navigation), Frequency domain, Computer science, Feature (linguistics), Time domain