System for assessment and prediction of the technical condition of power oil-filled transformer equipment of distribution networks using machine learning

Abstract

RELEVANCE the research is to develop a new system for assessing the technical condition of power oil-filled transformer equipment of distribution networks.OBJECT. To increase the accuracy of assessing the technical condition of power oil-filled transformer equipment (POTE) of distribution networks through the use of machine learning methods. Currently, an increase in the volume of analyzed information about the state of the management system of distribution networks leads to significant changes in the choice of data processing methods. The use of machine learning methods is associated both with the need to apply operational experience (in the form of expert assessments) and to obtain objective assessments of the condition of transformer equipment of distribution networks from instrumentation and sensors.METHODS. This work uses research methods such as mathematical modeling and the method of paired comparisons. As an example, we consider the oil-filled power transformer TMN-6300, its diagnostic parameters, external and operating parameters. The technical condition of the TMN-6300 transformer is assessed and a predictive model is created based on the existing monitoring system and machine learning methods, which make it possible to formalize expert knowledge and automate the process of data processing and analysis.RESULTS. A database has been created to assess and predict the technical condition of POTE of distribution network management systems. The algorithm for predicting the technical condition of POTE of the technical equipment in the form of an artificial neural network model was tested in the developed assessment system.CONCLUSION. The results of assessing and predicting the technical condition of POTE of the metering system of distribution networks obtained in this work prove the unconditional relationship between the parameters of the metering system and external, operating parameters. The data obtained as a result of modeling helps to increase the accuracy of forecasting the technical condition and determine the longterm prospects for the functioning of POTE the equipment management system, timely maintenance and repairs over the course of years and months.

Keywords

• control system
• power oil-filled transformer equipment
• distribution networks
• machine learning
• control methods
• assessment and forecasting of technical condition