A Comprehensive Diagnosis Method of Rolling Bearing Fault Based on CEEMDAN-DFA-Improved Wavelet Threshold Function...

Area of Science:

• Mechanical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Rolling bearings are critical components in rotating machinery.
• Effective fault diagnosis is essential for preventing catastrophic failures and ensuring operational reliability.
• Existing methods often struggle with noise interference and feature separability.

Purpose of the Study:

• To propose a comprehensive fault diagnosis method for rolling bearings.
• To enhance noise reduction and fault feature extraction capabilities.
• To improve the accuracy of fault identification in challenging conditions.

Main Methods:

• A novel denoising technique combining Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), Detrended Fluctuation Analysis (DFA), and improved wavelet thresholding.
• A fault feature extraction and identification strategy utilizing Quantum-behaved Particle Swarm Optimization (QPSO) to optimize Multiscale Permutation Entropy (MPE) parameters, coupled with Support Vector Machine (SVM).

Main Results:

• The CEEMDAN-DFA-improved wavelet thresholding method effectively reduces noise while preserving original signal characteristics.
• The QPSO-MPE-SVM approach successfully overcomes overlapping MPE values, enabling better separation of features for different fault types.
• Experimental validation achieved a fault identification accuracy of 95%.

Conclusions:

• The proposed comprehensive fault diagnosis method demonstrates superior performance in noise suppression and fault identification compared to existing techniques.
• The integration of advanced signal processing and machine learning algorithms offers a robust solution for rolling bearing condition monitoring.
• The high accuracy achieved highlights the potential of this method for practical industrial applications.