A NARX Model-Based Condition Monitoring Method for Rotor Systems

Area of Science:

• Mechanical Engineering
• Vibration Analysis
• Condition Monitoring

Background:

• Rotor systems are critical in many industrial applications.
• Early detection of faults in rotor systems is essential for preventing catastrophic failures.
• Traditional monitoring methods may not capture complex nonlinear behaviors.

Purpose of the Study:

• To develop a data-driven frequency domain analysis method for rotor systems.
• To propose a model-based index for fault feature detection using nonlinear models.
• To enhance condition monitoring capabilities for rotor systems.

Main Methods:

• Utilized Nonlinear Auto-Regressive with eXternal input (NARX) models based on vibration signals.
• Collected displacement vibration signals at multiple rotating speeds.
• Developed a Nonlinear Response Spectrum Function (NRSF)-based nonlinear fault index.
• Validated the method through experimental application on a misaligned rotor system.

Main Results:

• The NARX model successfully characterized the rotor system dynamics.
• The nonlinear fault index was calculated in a multi-frequency range.
• The index demonstrated a direct correlation with the severity of misalignment faults.
• Experimental results confirmed the effectiveness of the proposed method.

Conclusions:

• The data-driven frequency domain analysis method provides an effective approach for rotor system condition monitoring.
• The proposed nonlinear fault index is a reliable indicator of rotor system health and fault severity.
• This method facilitates early and accurate detection of faults in rotating machinery.