Bearing degradation assessment and remaining useful life estimation based on Kullback-Leibler divergence and Gaussian processes regression

•Propose a novel health indicator based on Kullback-Leibler divergence (KLD) method.•KLD value converted into the confidence value in the range of 0–1.•Estimation of residual life of bearing using the Gaussian Process Regression model.•The proposed method shows excellent performance on the experimental data set. Accurate monitoring of degradation in bearing is essential for preventing unexpected shutdown of a machinery system. This paper proposes a novel health degradation indicator for machineries, based on a Kullback-Leibler divergence. This would help assess degradation in rolling element bearings and make prediction on the remaining useful life. The confidence value is proposed as the bearing health degradation indicator for prognostics approach or for assessing the bearings' degradation trend. It is able to quantify different degradation stages (healthy, severe and failure stages) of rolling element bearings. Further, Gaussian Process Regression has been used to make the prognosis of degradation trend in bearings with a 95% confidence interval, remove outliers from confidence value, and estimate the remaining useful life of bearing. Experimental data on the test bearings were studied to demonstrate the proposed method's capabilities. It can assess an accurate degradation trend and health state of bearing.