Weibull Reliability Regression Model for Prediction of Bearing Remaining Useful Life
Journal
Computer Aided Chemical Engineering
Journal Volume
49
ISBN
9780323851596
Date Issued
2022-01-01
Author(s)
Abstract
In this work, a scenario-based approach that uses multiple Weibull Accelerated Failure Time Regression (WAFTR) models is proposed to predict the remaining useful life (RUL) of a benchmark bearing. The external features such as operational load and rotatory speed of the bearing are used to categorize the operational scenarios and a scenario-based WAFTR model is identified for each operational scenario by using the internal features extracted from the sampled horizontal and vertical vibration data. Therein, the Weibull parameters in each WAFTR model are expressed either in exponential or in linear form of these internal and external operational features. By using the mean squared error (MSE) as the performance measure of the prediction model, it is found that the proposed multiple WAFTR models approach can predict the RUL within 20% error.
Subjects
Bearing | Reliability function | Remaining Useful Life (RUL) | Weibull distribution
Type
book part