Cointegration Approach for Vibration-Based Misalignment Detection in Rotating Machinery Under Varying Load Conditions

Area of Science:

• Mechanical Engineering
• Condition Monitoring
• Signal Processing

Background:

• Shaft misalignment is a frequent issue in rotating machinery, posing challenges for traditional vibration-based diagnostic techniques, especially under variable operating loads.
• Existing methods struggle with reliable misalignment detection when load conditions fluctuate, necessitating advanced diagnostic frameworks.

Purpose of the Study:

• To develop and validate a new vibration-based framework for detecting shaft misalignment in rotating machinery.
• To address the challenge of reliable misalignment detection under varying load conditions using cointegration analysis.

Main Methods:

• A three-stage approach combining the Augmented Dickey-Fuller (ADF) test with cointegration analysis was employed.
• The method analyzes the stationarity of raw vibration signals, performs cointegration on time series, and re-evaluates stationarity on post-cointegrated data.
• Experimental validation was conducted on a test rig with a motor, gearbox, and hydraulic gear pump under healthy and misaligned states across various loads.

Main Results:

• The proposed cointegration-based framework successfully detected shaft misalignment despite variations in load conditions.
• The method demonstrated effectiveness in identifying the severity of misalignment.
• Residuals from the cointegration process showed high sensitivity to damage, proving valuable for condition monitoring.

Conclusions:

• The developed framework offers a robust solution for detecting shaft misalignment in rotating machinery under dynamic load conditions.
• Cointegration analysis and its residuals provide effective features for advanced vibration-based condition monitoring and fault diagnosis.