Comparison of Two Classification Machine Learning Models of Avionics Systems for Health Analysis

Abstract

Machine learning models are effectively applied to build digital twins of avionics systems under integrated system health management. This allows saving resources of manufacturers significantly. This paper describes a continuation of a research of machine methods application for avionics objects health analysis to solve classification problems under building digital twins in the integrated system health management. An additional model included a bigger quantity of features is considered in this part of the research. Comparison of classification results of models for two avionics objects had the different quantity of features shows that the same machine learning methods give different results. K-fold cross-validation was applied to get more accurate results. The application of the results will allows improving a base to build digital twins of avionics systems.