Data-Driven Method for Predicting Remaining Useful Life of Bearing Based on Bayesian Theory

Area of Science:

• Mechanical Engineering
• Reliability Engineering
• Data Science

Background:

• Bearings are critical industrial components; their failure significantly impacts equipment performance and incurs economic losses.
• Real-time estimation of remaining useful life (RUL) for bearings is essential for predictive maintenance and operational safety.
• Existing methods for RUL prediction often require complex modeling or extensive historical data.

Purpose of the Study:

• To propose a novel data-driven method for accurate Remaining Useful Life (RUL) prediction of bearings.
• To leverage Bayesian theory for robust state parameter estimation and life prediction.
• To validate the proposed method using open-source bearing datasets.

Main Methods:

• Extraction of time-domain features from bearing vibration signals.
• Data fusion to construct a Health Indicator (HI) and a bearing degradation state model.
• Establishment of a Bayesian model for state parameters and bearing life, updated via the Metropolis-Hastings algorithm.

Main Results:

• The proposed Bayesian method effectively predicts bearing RUL.
• Validation on the XJTU-SY bearing dataset demonstrated the accuracy of the RUL predictions.
• Comparison with existing methods confirmed the superior performance of the developed approach.

Conclusions:

• The data-driven Bayesian approach provides an accurate and reliable method for bearing RUL prediction.
• This method enhances predictive maintenance strategies for industrial machinery.
• The findings contribute to reducing economic losses associated with unexpected bearing failures.