A Model-Based Method for Remaining Useful Life Prediction of Machinery

TLDR

RUL prediction enables predictive maintenance and reduces costly unscheduled repairs, making it a hot and challenging research area. This study proposes a model‑based approach for predicting the remaining useful life of machinery. The approach constructs a weighted minimum quantization error health indicator from multiple features and then estimates RUL via maximum‑likelihood initialization followed by particle filtering, validated on vibration data from accelerated bearing degradation tests. The results demonstrate that the proposed method accurately predicts machinery RUL, confirming its effectiveness.

Abstract

Remaining useful life (RUL) prediction allows for predictive maintenance of machinery, thus reducing costly unscheduled maintenance. Therefore, RUL prediction of machinery appears to be a hot issue attracting more and more attention as well as being of great challenge. This paper proposes a model-based method for predicting RUL of machinery. The method includes two modules, i.e., indicator construction and RUL prediction. In the first module, a new health indicator named weighted minimum quantization error is constructed, which fuses mutual information from multiple features and properly correlates to the degradation processes of machinery. In the second module, model parameters are initialized using the maximum-likelihood estimation algorithm and RUL is predicted using a particle filtering-based algorithm. The proposed method is demonstrated using vibration signals from accelerated degradation tests of rolling element bearings. The prediction result identifies the effectiveness of the proposed method in predicting RUL of machinery.