Evaluation method of equipment health degree based on classification algorithm

Abstract

The accurate assessment of equipment health is the basis for the development of its equipment combat training and maintenance support countermeasures. In view of the problem that the health status of traditional equipment is difficult to scientifically assess, a comprehensive consideration of equipment life, use and maintenance, and the health status of equipment and its subsystems are proposed. Based on the equipment health evaluation model, combined with the characteristics of multi-classification of equipment health and unbalanced sample data, seven classification algorithms and ten sample balance methods are used to evaluate equipment health. Experiments show that the comprehensive use of Near-Miss, SMOTEENN, SMOTE Tomek sampling method and DT, GBDT, RF classification algorithms can effectively improve the indicators of equipment health classification, and provide an effective method for reasonable evaluation of equipment health status.

Keywords

• machine learning|classification algorithm|health assessment|unbalanced sample|evaluation index system