Weighted squared envelope diversity entropy as a nonlinear dynamic prognostic measure of rolling element bearing

Being a new nonlinear dynamic measure, diversity entropy (DE) is a promising parameter for prognostics of a rolling element bearing by quantifying the complexity of collected vibration signals. However, in real-life prognostic maintenance operations, distinctive signal components corresponding to a bearing fault get submerged under unnecessary random noise components. As a result, DE not only performs poorly in incipient detection of bearing faults but also fails to track fault growth in an efficient manner. In this paper, to overcome the aforementioned limitations of DE in bearing health prognosis, unwanted noise components associated with collected vibration signals are suppressed by weighting corresponding squared envelope. Due to the involvement of the weighted squared envelope, the proposed measure is termed as weighted square envelope based DE (WSEDE). Two different run-to-failure experimental datasets are used to validate the proposed measure. The results demonstrate that the proposed WSEDE not only overcomes the weaknesses of original DE in bearing health prognosis but also performs better than conventional fuzzy entropy (FE) and an advanced DE-based measure multiscale DE (MDE).

Automated summary

Diversity entropy (DE) is a new nonlinear dynamic measure that quantifies the complexity of vibration signals for prognostics of rolling element bearings. However, in real-life maintenance operations, the distinctive fault signals are often submerged under random noise components, leading to poor performance of DE in detecting and tracking bearing faults. To overcome these limitations, this paper proposes a weighted square envelope based DE (WSEDE) that suppresses unwanted noise components. Experimental results using two different datasets demonstrate that WSEDE outperforms the original DE and conventional fuzzy entropy (FE) as well as an advanced DE-based measure called multiscale DE (MDE).