Bearing health state assessment based on entropy weight-variation coefficient-TOPSIS combined method

Authors:
Open Access
Publisher website
Google Scholar
Abstract
To address the limited adaptability of fixed thresholds and the inherent drawbacks of single weighting methods in bearing state assessment, this paper proposes a bearing health evaluation model that integrates the entropy weight method, the variation coefficient method, and TOPSIS. Firstly, multi-dimensional time-domain features are extracted from bearing vibration signals to serve as health indicators. Then, indicator weights are determined by combining the entropy weight and variation coefficient methods, and Kendall’s coefficient of concordance is used to examine the agreement between the two weighting schemes. A comprehensive health score is subsequently computed using TOPSIS, and a dynamic health threshold is established based on the 3σ rule, enabling quantitative assessment of bearing health. Finally, the proposed approach is validated using the IEEE PHM 2012 bearing degradation dataset. Experimental results demonstrate that the method effectively tracks the full-lifecycle degradation of bearings from healthy operation through rapid deterioration to ultimate failure. The health score curve remains smooth during the healthy stage, the onset of degradation is accurately identified, and the overall evaluation performance surpasses that of single weighting methods. This paper provides valuable technical support for bearing health assessment and predictive maintenance.

This publication has 6 references indexed in Scilit: