Data-Driven Based Dissolved Gas Analysis Diagnosis Models for Oil-Immersed Power Transformers

Oil-immersed power transformers are essential components in power system, playing a curial role in maintaining the stability and reliable for the system. Degradation of insulation materials within the transformers can potentially lead faults such as Partial Discharge (PD), Low Energy Discharge (D1), High Energy Discharge (D2), Low Temperature Thermal Fault (T1), Medium Temperature Thermal Fault (T2), and High Temperature Thermal Fault (T3), all of which directly affect to the system. Dissolved gases, including ethylene (C₂H₄), acetylene (C₂H₂), methane (CH₄), ethane (C₂H₆), and hydrogen (H₂), are produced from the degradation. Each of these gases are associated with specific fault types, and their concentrations can be analyzed using Dissolved gas analysis (DGA) methods. In this study, transformer fault classifiers are developed using an ensemble model based on Kolmogorov-Arnold Networks (KANs). The ensemble KANs model demonstrates superior capability in identifying transformer faults.