Remaining Useful Life Prediction and Optimal Maintenance Time Determination for a Single Unit Using Isotonic Regression and Gamma Process Model

Degradation-based remaining useful life (RUL) prediction plays an important role in ensuring the reliability and safety of rotating machinery components. The accuracy of traditional models and methods is usually affected by the inevitable fluctuations in degradation signals. This paper proposes a dynamic RUL prediction and optimal maintenance time (OMT) determination approach using a Gamma process model. This approach can significantly reduce the effects of random fluctuations on the accuracy of RUL prediction, and facilitate the implementation of real-time condition-based maintenance. In particular, an isotonic regression based data preprocessing method, called pool-adjacent-violators algorithm, is first presented to smooth random fluctuations in degradation signals. Then, health stage identification is conducted by measuring the degradation gradient within a sliding window to characterize the degradation trend and identify the jump points. A Bayesian algorithm and a maximum likelihood estimation method are jointly utilized to update the model parameters and further predict the component's RUL. By considering both maintenance cost and failure risk of the component, an OMT determination method based on RUL prediction result is developed. A case study on rolling element bearings illustrates the superiority and effectiveness of the proposed approach in both RUL prediction and maintenance decision making.

Automated summary

This study presents a dynamic RUL prediction and OMT determination approach using a Gamma process model, which can significantly reduce the impact of random fluctuations on the accuracy of RUL prediction and facilitate real-time condition-based maintenance.