A Novel Health Indicator Based on Information Theory Features for Assessing Rotating Machinery Performance Degradation

Rotating machinery is used at length in a variety of industrial applications. The continuous monitoring of rotating machines is of excessive importance when it comes to prevent their catastrophic breakdown and the subsequent economic losses. This article proposes a novel health indicator to assess the performance degradation of two crucial rotating machinery components, namely rolling element bearings and gears. The potential of the proposed health indicator is demonstrated through experimental vibration signals acquired from several benchmark bearing and gear test rigs. The complexity measure, called as multiscale fuzzy entropy (MFEn), is extracted as a fault feature from the vibration signals. These MFEn feature vectors form probability distributions, the nature of which varies as the degradation in bearings or gears progresses. Then, the Jensen-Rényi divergence technique is applied, which discriminates the probability distribution of degraded multiscale entropy (MSE) feature vectors against the healthy MSE feature vectors to formulate the desired health indicator. Experimental results verify that the developed health indicator efficiently tracks the development of deterioration in rotary equipment and outperforms the conventional indicators, such as the root-mean-square value and kurtosis.