HKF-SVR Optimized by Krill Herd Algorithm for Coaxial Bearings Performance Degradation Prediction

Area of Science:

• Mechanical Engineering
• Signal Processing
• Machine Learning

Background:

• Industrial applications frequently use multiple bearings on a single shaft, resulting in mixed vibration signals.
• Accurate bearing performance degradation prediction is crucial for machinery health monitoring and maintenance.

Purpose of the Study:

• To develop and validate a robust method for predicting bearing performance degradation from mixed vibration signals.
• To enhance the accuracy and reliability of bearing life prediction in complex industrial settings.

Main Methods:

• Feature extraction from multi-domain vibration signals and fusion using Kernel Joint Approximate Diagonalization of Eigen-matrices (KJADE).
• Calculation of a performance degradation index based on scatter analysis of health-stage and monitored data.
• Optimization of Hybrid Kernel Function-Support Vector Regression (HKF-SVR) parameters using the Krill Herd (KH) algorithm.

Main Results:

• The proposed KJADE-enhanced HKF-SVR method demonstrated superior performance compared to traditional Back Propagation Neural Network (BPNN), Extreme Learning Machine (ELM), and Support Vector Regression (SVR).
• The method effectively extracts latent features from mixed signals, accurately reflecting bearing performance degradation.

Conclusions:

• The developed method offers a promising approach for accurate bearing performance degradation prediction in scenarios with mixed vibration signals.
• This technique has significant potential for application in the life prediction of coaxial bearings and similar industrial machinery.