A Rolling Bearing Fault Classification Scheme Based on k-Optimized Adaptive Local Iterative Filtering and Improved...

Area of Science:

• Mechanical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Rolling bearing health is critical for mechanical system operation.
• Bearing failures often result in nonlinear and non-stationary time series data.
• Accurate fault detection is essential for predictive maintenance.

Purpose of the Study:

• To propose a robust rolling bearing fault detection technique.
• To enhance the accuracy of fault characteristic extraction from complex time series.
• To enable automatic classification and identification of various fault types and severities.

Main Methods:

• Adaptive local iterative filtering (ALIF) optimized by permutation entropy (PE) for adaptive layer selection.
• A completely average coarse-graining method to enhance information extraction.
• Improved multiscale permutation entropy (improved MPE) for analyzing intrinsic mode functions (IMFs).
• Backpropagation (BP) neural network for fault identification using extracted features.

Main Results:

• The improved MPE effectively extracts deep information from time series, yielding stable entropy values.
• k-optimized ALIF decomposes rolling bearing data into IMFs for analysis.
• The proposed method accurately extracts fault features and classifies different fault modes and degrees.
• Simulation and experimental results validate the effectiveness of the technique.

Conclusions:

• The integrated approach of k-optimized ALIF, improved MPE, and BP neural network offers effective rolling bearing fault detection.
• This method demonstrates significant potential for real-world applications in bearing fault identification and diagnosis.
• The technique successfully addresses the challenges posed by nonlinear and non-stationary fault signals.