Data Driven Condition Monitoring Model of Power Transformer for Diagnosing Incipient Faults in Smart City Network

Abstract

Power transformers are an essential part of smart city infrastructure and are also one of the most expensive components in this type of infrastructure. The insulation of the transformers is made from mineral oil and cellulose, both of which can deteriorate as a result of multiple stresses (electrical/mechani- cal/thermal/chemical). This paper presents a new computational model for mak- ing asset decisions that incorporates various significant variables, including dis- solved gas analysis (DGA), water content, furan levels, interfacial tension, and degree of polymerization. Different stresses on its insulating structure are also analyzed using contour plots and surface viewers. To test and validate this expert model, 200 transformer’s data driven analysis is used. Gas ratio techniques, the Duval Triangle technique, the degree of polymerization and furans-based paper deterioration, the moisture and IFT-based insulation degradation, and other dis- solved gas analysis-based diagnostic procedures are used and its shows the higher accuracy analysis and efficiency for incipient faults diagnosis and analysis with AI based computational intelligence.