APPLICATION OF THE DECISION TREE ALGORITHM TO EVALUATE THE RESULTS OF CHROMATOGRAPHIC ANALYSIS OF TRANSFORMER OIL

Abstract

Power oil-filled transformer is the main link in the process of conversion and transmission of electrical energy in electrical networks at almost all voltage classes. At the moment in our country more than 50% of them are operated with significant excess of service life, which increases the requirements to their proper technical control and full-fledged diagnostics of all its main components.
 
 The aim of the research is to consider the ways of simplification and automation of the process of technical diagnostics of power oil-filled equipment by means of application of artificial intelligence methods, namely the decision tree algorithm for evaluation of the results of chromatographic analysis of transformer oil.
 
 Materials and Methods. As input data, the results of chromatographic analysis of dis-solved gases in transformer oil conducted from December 11, 2009 to December 12, 2020 for two power transformers of voltage class 110 kV installed in the power system of one of the regions of our country were considered and analyzed. As a software application used for diagnosing the technical condition of a power transformer using artificial intelligence methods, we selected the free application Deductor Academic 5.3 Build 0.46.
 
 Research results. The article considers the application of a promising method (decision tree algorithm) of interpretation of data obtained as a result of chromatographic analysis of dissolved gases in transformer oil. The analysis data were processed by means of artificial intelligence methods, the result of which was the reliability and accuracy of determining the technical condition of the power oil-filled transformer.
 
 Conclusions. According to the results of the study, a decision tree algorithm is proposed for the implementation of artificial intelligence in solving the problem of power transformer diagnostics using chromatographic analysis results.