Dual-Domain Impulse Complexity Index-Guided Projection Iterative-Methods-Based Optimizer-Feature Mode Decomposition...

Area of Science:

• Mechanical Engineering
• Signal Processing
• Condition Monitoring

Background:

• Bearings are critical industrial components susceptible to noise interference, reducing signal-to-noise ratio (SNR) in vibration analysis.
• Traditional methods struggle to extract bearing fault information effectively under low SNR conditions.
• Feature Mode Decomposition (FMD) shows promise but is limited by parameter sensitivity and manual settings.

Purpose of the Study:

• To develop an adaptive parameter optimization method for Feature Mode Decomposition (FMD) to improve bearing fault detection.
• To enhance the extraction of fault impact components from vibration signals with low SNR.
• To improve the stability and reliability of bearing fault diagnosis in noisy industrial environments.

Main Methods:

• Proposed a Dual-domain Impulse Complexity Index (DICI) combining time-domain and frequency-domain characteristics for FMD parameter evaluation.
• Employed a projection-iterative-methods-based optimizer (PIMO) for adaptive optimization of FMD parameters.
• Utilized the Fault Frequency Correlation (FFC) criterion for sensitive component selection and envelope spectra analysis for fault mode recognition.

Main Results:

• The proposed DICI-PIMO method demonstrated superior performance in identifying bearing faults compared to existing methods under low SNR conditions.
• Adaptive parameter optimization significantly improved the stability and effectiveness of FMD.
• Successful verification using both simulated and real industrial vibration signals.

Conclusions:

• The DICI-PIMO approach offers a robust and adaptive solution for bearing fault diagnosis in challenging, high-noise industrial environments.
• This method enhances the capability of vibration signal analysis for effective condition monitoring.
• The findings suggest a significant advancement in detecting bearing failures with low SNR.