Control System for Assessing Reliability Functioning of the Complex Radio Electronic Equipment Using Machine Learning Methods

We propose a novel application of machine learning methods to predict the state of complex structured radio electronic systems. The model for collecting data about the functioning of a radio electronic system based on an integrated monitoring system of its components states is proposed and justified. The proposed model made possible to adapt one of the most widespread methods of machine learning, gradient boosting, on a sample of historical data, from the General Designer stand and embedded control system, to solve the problem of forecasting the state of the whole system. Operating experience showed that component failures are extremely rare. Therefore, the distribution of statistics of failures of the past break-in elements was considered according to Poisson's law, and the elements introduced for the first time according to Weibull's law with different coefficients, the prediction of such coefficients based on historic data is just the result of the machine learning algorithm.