Refined Composite Multiscale Fuzzy Dispersion Entropy and Its Applications to Bearing Fault Diagnosis

Abstract

Rotary machines often exhibit nonlinear behavior due to factors such as nonlinear stiffness, damping, friction, coupling effects, and defects. Consequently, their vibration signals display nonlinear characteristics. Entropy techniques prove to be effective in detecting these nonlinear dynamic characteristics. Recently, an approach called fuzzy dispersion entropy (DE–FDE) was introduced to quantify the uncertainty of time series. FDE, rooted in dispersion patterns and fuzzy set theory, addresses the sensitivity of DE to its parameters. However, FDE does not adequately account for the presence of multiple time scales inherent in signals. To address this limitation, the concept of multiscale fuzzy dispersion entropy (MFDE) was developed to capture the dynamical variability of time series across various scales of complexity. Compared to multiscale DE (MDE), MFDE exhibits reduced sensitivity to noise and higher stability. In order to enhance the stability of MFDE, we propose a refined composite MFDE (RCMFDE). In comparison with MFDE, MDE, and RCMDE, RCMFDE’s performance is assessed using synthetic signals and three real bearing datasets. The results consistently demonstrate the superiority of RCMFDE in detecting various patterns within synthetic and real bearing fault data. Importantly, classifiers built upon RCMFDE achieve notably high accuracy values for bearing fault diagnosis applications, outperforming classifiers based on refined composite multiscale dispersion and sample entropy methods.

Benefits of Zero-Phase or Linear Phase Filters to Design Multiscale Entropy: Theory and Application

“…In [84], the authors suggested using the refined composite multiscale DE and applying it to biomedical signals. More recently, they considered the refined composite generalized MDE based on the mean, variance, and skewness [85], and they proposed the multiscale fuzzy dispersion entropy (MFDE) and refined composite MFDE (RCMFDE) [86].…”

Fault Diagnosis Method for Rolling Bearings Based on Grey Relation Degree

“…They are required to ensure high-precision continuous operation in various complex and harsh environments. However, continuous operation under high load and variable working conditions can easily cause various faults in rolling bearings, leading to equipment shutdown and affecting the production process [ 1 , 2 , 3 ]. Therefore, how to effectively extract fault features from the full life cycle data of rolling bearings and accurately diagnose different fault states and degrees is of great practical significance for ensuring the safe and stable operation of mechanical equipment.…”