Rolling Bearing Fault Diagnosis Based on Refined Composite Multi-Scale Approximate Entropy and Optimized...

Area of Science:

• Mechanical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Early fault diagnosis in rolling bearings is crucial for mechanical system reliability and safety.
• Strong background noise often masks early fault signatures, complicating diagnosis.
• Existing methods may struggle with noise interference and feature extraction accuracy.

Purpose of the Study:

• To propose a robust rolling bearing early fault diagnosis method.
• To enhance the accuracy of fault recognition in noisy environments.
• To improve the reliability and safety of mechanical fault diagnosis systems.

Main Methods:

• Signal decomposition using composite ensemble intrinsic time-scale decomposition with adaptive noise (CEITDAN).
• Complexity analysis of vibration signals via refined composite multi-scale approximate entropy (RCMAE).
• Pattern recognition using an improved coyote optimization algorithm-based probabilistic neural network (ICOA-PNN).

Main Results:

• The proposed RCMAE method effectively analyzes signal complexity.
• The ICOA-PNN classifier achieved high recognition accuracy in fault diagnosis.
• Experimental validation confirmed the method's feasibility and effectiveness for early fault detection.

Conclusions:

• The integrated CEITDAN-RCMAE-ICOA-PNN approach offers a promising solution for rolling bearing early fault diagnosis.
• This method demonstrates superior performance in handling noisy vibration signals.
• The findings contribute to enhanced safety and operational efficiency in mechanical systems.