Gearbox Rolling Bearing Fault Diagnosis Based on Autocorrelation Envelope and Adaptive MED

Minimum entropy deconvolution (MED) is a popular algorithm in recent years, extensively applied in feature extraction and fault diagnosis of components such as gearboxes and bearings. However, in the actual computational process, the parameter settings of the MED inverse filter are highly sensitive to the extraction results. To address this issue, an optimized method was firstly proposed for extracting the fault characteristics of rolling bearings using MED. This method takes into account the energy proportion of feature frequencies under different lengths of inverse filters during the MED computation process, thereby determining the optimal parameters for the inverse filter. Additionally, the self-correlation of envelope signals is utilized to further enhance the weak fault characteristic signals of rolling bearings. By integrating self-correlated envelopes with the optimized MED method, a novel method for feature extraction and fault diagnosis of gearbox rolling bearings has been developed. Simulations and tests have verified that this method can effectively enhance the characteristic signals related to bearing faults, and the optimized MED method is significantly superior to the traditional MED and other related bearing signal processing methods. Notably, the self-correlated envelope, due to its ability to significantly enhance impulse components and its excellent denoising characteristics, shows more prominent results in the actual diagnosis of gearbox bearing faults.