Hybrid Gaussian process regression with temporal feature extraction for partially interpretable remaining useful life...

Area of Science:

• * Intelligent Manufacturing and Industry 4.0
• * Machine Learning for Predictive Maintenance

Background:

• * Remaining Useful Life (RUL) estimation is critical for intelligent manufacturing and Industry 4.0.
• * Existing RUL models often struggle with interpretability and robust uncertainty quantification.
• * Accurate RUL prediction is essential for optimizing manufacturing processes and reducing downtime.

Purpose of the Study:

• * To introduce an adapted Gaussian Process Regression (GPR) model for RUL interval prediction.
• * To address the challenges of interpretability and uncertainty modeling in RUL estimation.
• * To enhance the accuracy and transparency of RUL predictions in manufacturing.

Main Methods:

• * Utilized an adapted Gaussian Process Regression (GPR) model for RUL interval prediction.
• * Coupled GPR with deep adaptive learning-enhanced AI process models to capture complex patterns.
• * Incorporated feature significance evaluation for transparent decision-making.

Main Results:

• * The adapted GPR model effectively predicts confidence intervals for RUL.
• * The approach successfully captures intricate time-series patterns and dynamic manufacturing behaviors.
• * Feature significance evaluation provided interpretable insights into RUL prediction drivers.

Conclusions:

• * The proposed GPR model offers a structured approach to uncertainty modeling in RUL prediction.
• * This method enhances the accuracy and interpretability of RUL estimations in manufacturing.
• * The findings contribute to more robust process development and management through reliable RUL insights.