13 / 2025-04-11 15:41:21

A suitable range selection method for MED filter length and its application

Rolling bearing,fault diagnosis,Minimal Entropy Deconvolution (MED),filter length

Special Sessions > SSL 17 Advanced intelligent methodologies for fault detection, diagnosis and RUL prediction of aerospace systems

Minimal Entropy Deconvolution (MED) can mitigate the influence of complex transmission paths, but its effectiveness depends on the choice of the filter length. In this study, we investigate the relationship between the signal characteristics and the filter performance of MED as the filter length increases, aiming to determine a suitable range of the filter length. First, we construct various indices based on four different characteristics: periodicity, kurtosis, cyclostationarity, and entropy. Then two indices that can reflect the filter effect are designed to discuss the correlations between the characteristic indices and the filter performance. Analysis of both simulation data and measured signals from an aircraft accessory gearbox reveals that as the filter length increases, the filter performance can be categorized into four distinct zones: the initial barren zone, the abundant zone, the fluctuation zone, and the final barren zone. By combining the jump points of characteristic indexes, we can accurately delineate the boundary of each zone. Selecting the filter length within the abundant zone will result in feature-rich results, making the MED effective for fault diagnosis.