Data Decomposition Techniques with Multi-Scale Permutation Entropy Calculations for Bearing Fault Diagnosis

Area of Science:

• Mechanical Engineering
• Signal Processing
• Fault Diagnosis

Background:

• Rolling element bearings (REBs) are critical components in machinery.
• Early fault detection in REBs is essential for preventing catastrophic failures and ensuring operational reliability.
• Traditional methods may struggle with complex vibration signal characteristics.

Purpose of the Study:

• To develop and validate an integrated LMD-MPE technique for enhanced REB fault diagnosis.
• To extract robust, damage-sensitive nonlinear features from vibration signals.
• To compare the proposed LMD-MPE method against existing techniques.

Main Methods:

• Local Mean Decomposition (LMD) was applied to decompose vibration signals into product functions (PFs).
• Multi-Scale Permutation Entropy (MPE) was used to calculate statistical permutation entropy from the PFs.
• The extracted nonlinear features were used to assess and classify the condition of REBs.

Main Results:

• The LMD-MPE approach successfully extracted reliable and damage-sensitive features from REB vibration data.
• Comparative analysis demonstrated that LMD-MPE yields more vigorous outcomes than conventional methods.
• The technique effectively distinguishes between healthy and damaged REB conditions.

Conclusions:

• The integrated LMD-MPE method offers a powerful tool for accurate and sensitive REB fault diagnosis.
• This approach enhances the ability to analyze bearing conditions from vibration measurements.
• The proposed technique shows significant potential for improving machinery health monitoring systems.