Composite Multiscale Transition Permutation Entropy-Based Fault Diagnosis of Bearings

Area of Science:

• Signal Processing
• Machine Learning
• Mechanical Engineering

Background:

• Transition permutation entropy (TPE) quantifies signal complexity but is limited to a single scale.
• Existing multiscale methods (MTPE) suffer from data loss during coarse-graining, reducing accuracy.
• Defects in multiscale analysis hinder effective fault feature extraction from vibration signals.

Purpose of the Study:

• To introduce a composite multiscale transition permutation entropy (CMTPE) method for improved fault diagnosis.
• To address the limitations of existing multiscale entropy methods, particularly data loss and accuracy.
• To enhance the robustness, noise resistance, and feature extraction capabilities for bearing fault diagnosis.

Main Methods:

• Developed a composite multiscale transition permutation entropy (CMTPE) approach.
• Applied CMTPE for feature extraction from vibration signals.
• Integrated CMTPE with an extreme learning machine (ELM) for a fault diagnosis strategy.

Main Results:

• The proposed CMTPE method overcomes coarse-graining defects, preserving key signal information.
• CMTPE demonstrates superior performance in feature extraction and entropy estimation accuracy.
• The CMTPE-ELM strategy shows enhanced robustness, accuracy, and stability in bearing fault diagnosis compared to other methods.

Conclusions:

• The composite multiscale transition permutation entropy (CMTPE) method significantly improves upon existing techniques for analyzing signal complexity.
• The CMTPE-ELM fault diagnosis strategy offers a robust and accurate solution for bearing fault identification.
• This approach provides a stable and reliable method for complex fault diagnosis in mechanical systems.