Performance Degradation Prediction Based on a Gaussian Mixture Model and Optimized Support Vector Regression for an...

Area of Science:

• Aerospace Engineering
• Mechanical Engineering
• Signal Processing

Background:

• Effective aviation pump health management and condition-based maintenance rely on accurate performance degradation prediction.
• Existing feature extraction methods may not fully utilize intrinsic mode function (IMF) information for degradation assessment.

Purpose of the Study:

• To propose a novel approach combining Gaussian mixture model (GMM) and optimized support vector regression (SVR) for predicting aviation pump degradation.
• To develop a quantitative degradation index (DI) based on GMM and self-information quantity for assessing pump health.
• To enhance SVR prediction accuracy through phase space reconstruction and a hybrid particle swarm optimization-grid search (PSO-GS) parameter tuning method.

Main Methods:

• Multi-domain features were extracted from selected IMF components of pump outlet pressure signals.
• Principal Component Analysis (PCA) was employed for sensitive degradation feature selection.
• A degradation index (DI) was defined using GMM and self-information quantity to quantify pump degradation.
• SVR model inputs were determined using phase space reconstruction, and parameters were optimized via PSO-GS.
• The SVR model was trained using both online and historical aviation pump data.

Main Results:

• The proposed DI effectively identifies and tracks the current deterioration stage of aviation pumps.
• The optimized SVR model demonstrated superior prediction accuracy compared to existing methods.
• Validation using full life cycle data from an aviation pump test rig confirmed the approach's effectiveness.

Conclusions:

• The developed GMM-SVR approach offers a robust and accurate method for aviation pump degradation prediction.
• The quantitative DI provides a valuable metric for real-time health monitoring and maintenance scheduling.
• This methodology contributes to advancing aviation pump health management and condition-based maintenance strategies.