Identification of mechanical compound-fault based on the improved parameter-adaptive variational mode decomposition

Area of Science:

• Signal Processing
• Mechanical Engineering
• Data Analysis

Background:

• Traditional variational mode decomposition (VMD) is hindered by predefined parameters, impacting industrial applications.
• Existing parameter-adaptive VMD methods struggle with optimal mode selection and compound-fault diagnosis.

Purpose of the Study:

• To propose an improved parameter-adaptive VMD (IPAVMD) for enhanced fault diagnosis in industrial data.
• To address limitations in mode number selection and improve compound-fault detection capabilities.

Main Methods:

• Developed a novel 'ensemble kurtosis' index combining kurtosis and envelope spectrum kurtosis for cyclostationary and impulsive feature detection.
• Enhanced the grasshopper optimization algorithm's objective function using the mean ensemble kurtosis across all modes.
• Implemented an iterative algorithm to ensure comprehensive extraction of potential fault information.

Main Results:

• The proposed IPAVMD demonstrates superior performance compared to traditional parameter-adaptive VMD.
• Successfully identified compound faults, outperforming existing methods.
• Validated through simulated signals and a real-world dataset from locomotive axle box bearings.

Conclusions:

• IPAVMD offers a significant advancement in VMD for industrial fault diagnosis, particularly for complex compound faults.
• The method effectively extracts fault signatures and improves diagnostic accuracy in challenging industrial environments.