Weighted Squared Envelope Dispersion Entropy as a Nonlinear Measure for Dynamic Health Monitoring of Rotating Machineries

Being a new non-linear measure, dispersion entropy (DisE) is a promising parameter for dynamic health monitoring of rotating machineries. However, under heavy noisy scenario, transient impulses corresponding to a rotating machinery fault gets submerged under the noise components present in the collected vibration signal. As a result, DisE not only fails to detect the presence of a fault at the earliest stage of inception but also performs poorly in monitoring the growth of the fault. Aiming at overcoming the limitations of DisE in dynamic health monitoring of rotating machineries, in this paper, impulses corresponding to a fault is extracted by suppressing the unnecessary noise component by weighting the squared envelope of the collected vibration signal. Due to the application of weighted squared envelope in calculating the DispE, the proposed measure is termed as weighted squared envelope dispersion entropy (WSEDisE). Effectiveness of WSEDisE in dynamic health monitoring of rotating machineries is verified by an experimental run to failure bearing vibration data. Experimental results show that WSEDisE not only overcomes the weaknesses of original DisE but also demonstrates better performance than conventional entropy-based methods such as permutation entropy (PE) and advanced DisE based method namely multiscale DisE (MDisE).