Refined Composite Multiscale Fuzzy Dispersion Entropy and Its Applications to Bearing Fault Diagnosis

Area of Science:

• Mechanical Engineering
• Signal Processing
• Nonlinear Dynamics

Background:

• Rotary machines exhibit nonlinear behavior due to various factors, leading to complex vibration signals.
• Entropy techniques are effective for detecting nonlinear dynamics, but existing methods like fuzzy dispersion entropy (FDE) and multiscale fuzzy dispersion entropy (MFDE) have limitations.
• MFDE offers improved stability and noise reduction compared to multiscale dispersion entropy (MDE), but further enhancements are needed.

Purpose of the Study:

• To introduce a refined composite multiscale fuzzy dispersion entropy (RCMFDE) method to enhance the stability and diagnostic accuracy of bearing fault detection.
• To evaluate the performance of RCMFDE against existing methods like MFDE, MDE, and refined composite multiscale dispersion entropy (RCMDE).
• To demonstrate the effectiveness of RCMFDE in classifying bearing faults using real-world datasets.

Main Methods:

• Development of the refined composite multiscale fuzzy dispersion entropy (RCMFDE) algorithm.
• Assessment of RCMFDE using synthetic time series data.
• Validation of RCMFDE on three distinct real-world bearing datasets, comparing its performance with MFDE, MDE, and RCMDE.
• Implementation of classifiers based on RCMFDE for bearing fault diagnosis.

Main Results:

• RCMFDE demonstrates superior performance in detecting fault patterns in both synthetic and real bearing data compared to MFDE, MDE, and RCMDE.
• Classifiers utilizing RCMFDE achieve significantly higher accuracy in bearing fault diagnosis.
• RCMFDE outperforms classifiers based on refined composite multiscale dispersion and sample entropy methods.

Conclusions:

• RCMFDE is a robust and effective method for analyzing nonlinear vibration signals from rotary machines.
• The proposed RCMFDE significantly improves the accuracy and stability of bearing fault diagnosis.
• RCMFDE shows great potential for practical applications in condition monitoring and fault diagnosis of rotating machinery.