A crayfish-optimized wavelet filter and its application to fault diagnosis

TL;DR

This paper introduces a crayfish optimization algorithm to adaptively tune wavelet filters for fault diagnosis in industrial machinery, improving the extraction of fault-related frequency components amidst noise.

Contribution

It presents a novel crayfish optimization-based method for adaptive wavelet filter parameter tuning, enhancing fault feature extraction in noisy industrial signals.

Findings

Outperforms existing methods in fault frequency extraction

Effective in low SNR environments

Demonstrated on multiple industrial cases

Abstract

Industrial machine fault diagnosis ensures the reliability and functionality of the system, but identifying informative frequency bands in vibration signals can be challenging due to low signal-to-noise ratio (SNR), background noise, and random interferences. The wavelet filter is commonly used for this purpose, but its parameters are crucial for locating the informative frequency band to extract repetitive transients. This study utilizes a crayfish optimization algorithm (COA) to optimize the wavelet filter adaptively for extracting fault characteristics. COA uses correlated kurtosis (CK) as a fitness function while addressing issues related to inaccurate CK period through an updation process. The proposed methodology is applied to different industrial cases and compared with existing methods, demonstrating its superiority in extracting informative frequencies.