A Hybrid Diagnostic Framework with Compensation Algorithms for Inherent Rotor Faults Using Rotor Experiments.

Area of Science:

• Engineering
• Mechanical Vibrations
• Prognostics and Health Management (PHM)

Background:

• Rotor-bearing systems in engineering applications suffer from inherent faults like imbalance and shaft-bow.
• Early identification of these faults is crucial for accurate prognostics and health management (PHM).
• Undetected inherent faults can lead to significant deviations in fault diagnosis and compromise maintenance strategies.

Purpose of the Study:

• To develop a hybrid diagnostic methodology for enhanced rotor fault diagnosis.
• To improve the simultaneous identification and parameter estimation of imbalance and shaft-bow faults.
• To address discrepancies between model predictions and experimental results caused by inherent faults.

Main Methods:

• Integration of a physics-based model with neural network techniques.
• Demonstration using a Jeffcott rotor model with simultaneous disk imbalance and shaft-bow.
• Development and experimental verification of algorithms for inherent fault identification and compensation.

Main Results:

• The hybrid methodology demonstrated superior capability for simultaneous fault identification.
• Discrepancies between model predictions and experimental data were observed due to inherent faults.
• Compensation for inherent faults significantly improved the accuracy of simultaneous diagnosis and parameter estimation.

Conclusions:

• The proposed hybrid methodology effectively enhances rotor fault diagnosis.
• Inherent fault identification and compensation are critical for accurate prognostic assessments.
• The methodology is suitable for real-time monitoring systems in industrial rotating machinery.