Interpretable equipment health state assessment method based on evidential reasoning with statistical feature

To ensure the proper operation of equipment, it is essential to assess its health state. The variables involved in equipment monitoring are typically high-dimensional and correlated; however, most assessment methods require these indicators to be independent. Additionally, the interpretability of the modeling process must be considered to ensure that the assessment results are credible and comprehensible. In this paper, we propose an interpretable equipment health state assessment model based on evidential reasoning (ER) with statistical features, named ISF-ER. This model establishes an interpretable mapping relationship between input and output spaces. The transformation from data to belief distribution is conducted based on reference values. Subsequently, a new interpretable evidence expression is constructed using statistics and control limits derived from the feature extraction of principal component analysis (PCA). Moreover, knowledge of the underlying mechanisms is integrated to establish the indicator relation matrix and determine the evidence weights. Finally, the evidence elements are fused based on ER to obtain the assessment results. A case study on the health state assessment of an inertial measurement unit (IMU) is presented to validate the effectiveness of the proposed model.