Optimal frequency band selection using blind and targeted features for spectral coherence-based bearing diagnostics:...

Area of Science:

• Mechanical Engineering
• Signal Processing
• Condition Monitoring

Background:

• Envelope spectrum analysis is crucial for bearing fault diagnosis.
• Selecting informative spectral frequency bands from spectral coherence is challenging.
• Existing blind and targeted features have limitations in discriminating optimal bands.

Purpose of the Study:

• To introduce a novel targeted feature for quantifying signal-to-noise ratio in spectral coherence.
• To develop a method for identifying the optimal spectral frequency band for bearing diagnostics.
• To compare the proposed method with existing approaches for improved fault detection.

Main Methods:

• A new targeted feature is proposed to assess signal-to-noise ratio in narrow frequency bands of spectral coherence.
• A method is developed using this feature to select the optimal spectral frequency band.
• The method's efficiency is validated against blind and targeted feature-based methods using simulated and experimental bearing data.

Main Results:

• The developed targeted feature effectively evaluates bearing failure information in the cyclic frequency domain.
• The presented approach accurately discriminates failure-related spectral frequency bands.
• The method demonstrates superior performance in detecting various bearing faults compared to state-of-the-art features.

Conclusions:

• The novel targeted feature and associated method provide an efficient and accurate approach for bearing fault diagnosis.
• The study highlights the advantages of the proposed method in identifying defect-sensitive frequency bands.
• The findings guide the application of feature-based methods for improved condition monitoring of bearings.