Fault Diagnosing of Cycloidal Gear Reducer Using Statistical Features of Vibration Signal and Multifractal Spectra

Area of Science:

• Mechanical Engineering
• Condition Monitoring
• Vibration Analysis

Background:

• Cycloidal gearboxes are susceptible to sliding sleeve damage, potentially leading to catastrophic failure.
• Early detection of this damage is critical for preventing extensive damage and ensuring operational reliability.
• Existing diagnostic methods may not be sufficiently sensitive to this specific type of gear damage.

Purpose of the Study:

• To develop and evaluate a novel method for diagnosing simulated cycloidal gear damage.
• To investigate the effectiveness of multifractal analysis for detecting damage to sliding sleeves.
• To identify robust signal features for early fault detection in cycloidal gearboxes.

Main Methods:

• Simulated damage by removing sliding sleeves from cycloidal gearbox pins.
• Recorded signals from torque, rotational speed, and vibration sensors under varying loads and speeds.
• Performed frequency analysis and calculated higher-order statistical moments (e.g., kurtosis).
• Applied multifractal analysis using the wavelet leader method (WLMF) to vibration signals.

Main Results:

• Frequency analysis proved insufficient for detecting the simulated damage due to rotational speed fluctuations.
• Higher-order statistical moments like kurtosis showed some sensitivity to the damage.
• Multifractal analysis, specifically log-cumulants of the multifractal spectrum, emerged as sensitive features for damage detection.

Conclusions:

• Multifractal analysis of vibration signals offers a promising approach for early detection of cycloidal gearbox sliding sleeve damage.
• Log-cumulants derived from wavelet leader multifractal analysis are effective diagnostic features.
• This method provides a more sensitive alternative to traditional frequency-based techniques for this specific fault.