Low-Speed Fault Diagnosis of Machinery Based on Improved Spectral Amplitude Modulation with Advanced Denoising...

Area of Science:

• Mechanical Engineering
• Condition Monitoring
• Signal Processing

Background:

• Offshore wind turbine bearings face significant background noise, masking weak fault signatures.
• Conventional spectral amplitude modulation methods suffer from degraded fault detection accuracy in noisy conditions.

Purpose of the Study:

• To propose a novel hybrid diagnosis method for low-speed bearing fault detection in offshore wind turbines.
• To address the challenge of noise interference in identifying incipient and weak fault characteristics.

Main Methods:

• Integrating fast spectral coherence and spectral amplitude modulation for enhanced signal processing.
• Implementing spectral amplitude modulation on raw vibration signals.
• Employing fast spectral coherence to eliminate noise and reconstruct enhanced envelope spectra.
• Utilizing squaring transformation and amplitude normalization to amplify weak fault signatures.

Main Results:

• The hybrid method effectively eliminates complex noise interference.
• Enhanced envelope spectra with amplified fault signatures were successfully reconstructed.
• Validated through simulated signals with varying signal-to-noise ratios and experimental data of bearing faults at 60 RPM.
• Demonstrated significant improvement over traditional diagnosis methods.

Conclusions:

• The proposed hybrid method offers a robust and reliable solution for incipient fault monitoring.
• It significantly remedies the limitations of traditional diagnosis methods in noisy environments.
• Provides a valuable technical solution for health assessment of low-speed bearings in offshore wind farms.