Development and Comparison of Methods for Predicting the Technical Condition of Alternators Powered By Gas Piston Engines Based on the Analysis of an Array of Historical Data

Obviously, the effectiveness of diagnostics increases significantly when the task of predicting changes in the state of an object at future points in time is solved. This is due to the fact that the ability to predict unmanageable aspects of the operation of an object before making a decision to put it into repair allows you to make an optimal choice of the timing and scope of repair work. Currently, there are several approaches to solving the problem of forecasting the technical condition, including methods based on the analysis of historical data, the advantage of which is to take into account the influence of external factors such as the interval between repairs, the quality of maintenance and repair, sudden changes in loads, etc. The purpose of this study is to develop and compare models for predicting the technical condition of alternators powered by a gas piston engine (gas piston units) based on an array of historical data on the technical condition. Adaptive neuro-fuzzy network and regression models of time series were used in the development of models. Conclusions are drawn about the highest accuracy of the regression model of the Gaussian process to describe the process of changing the technical condition of gas piston units. The proposed methods made it possible to accurately predict changes in the technical condition of gas piston units, taking into account the influence of external factors. The results obtained can be used as a basis for the creation of predictive maintenance systems for gas piston units.