Degradation assessment and trend prediction of rolling bearing based on MD-ESN

Abstract

It is of significance to monitor the operating status of rolling bearings. In order to obtain degradation of bearing performance timely and predict its trend, this paper proposed a bearing health state assessment method based on Mahalanobis distance metric and a prediction method based on echo state network. By designing degradation experiment and applying the proposed method to analyze the data, the performance degradation curve of the bearing and its prediction curve were obtained, and the experimental results verified the effectiveness of the method.

Predicting RUL of Bearings in the Pumps of a Critical Industrial Air Treatment System

“…There is a vast literature about statistical models, including linear regression, process models, Bayesian filtering, covariate-based hazard, and Markov models (Jin, Que, Sun, Guo, & Qiao, 2019). (Zhang & Li, 2014) used accelerated degradation testing to determine the health condition and RUL of bearings, making use of linear Wiener process models to perform reliability analysis under different stress levels, while (Zhao, Tang, & Tan, 2016) proposed a regressionbased solution after the application of Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) on a time-frequency representation of vibration signals.…”

“…Feature extraction and selection are important steps in capturing the bearing health status in operation (Nguyen et al, 2018;Kim et al, 2016;Soualhi et al, 2014;Rac ¸ões et al, 2019;Lv et al, 2018;Zhang & Li, 2014;Zhao et al, 2016). Various feature extraction methods, that typically consist of time-and spectral-domain analysis of bearing vibration sig-Figure 2.…”