Multi-point grounding fault diagnosis and temperature field coupling analysis of oil-immersed transformer core based on finite element simulation

In this article, a fault diagnosis method is presented for identifying multi-point grounding faults in transformer cores. Based on the finite element method (FEM), the method aims to achieve non-invasive localization of these faults, thereby minimizing loss attributed to troubleshooting and maintenance downtime. Firstly, as a prerequisite for solving the coupled temperature field, an applied circuit method is employed to ascertain the internal loss of the transformer. This determination relies on investigating the influence of internal loss in the oil-immersed transformers for diagnosing under rated operating conditions. Secondly, the internal temperature distribution within the transformer and the condition of the transformer oil are comprehensively analyzed and validated by using an electromagnetic-fluid-solid coupled finite element simulation. Lastly, relying on the simulation outcomes under fault-free conditions, a fault model of the transformer core is established. Subsequently, the output data of the model are classified and localized using deep belief networks. This approach resolves the problems associated with masking transformer fault data signals and acquiring data. The results indicate that the temperature data obtained from the temperature field simulation exhibit an error of less than 2.6% when compared to actual values. The fault localization accuracy for multiple grounding points in the iron core of oil-immersed transformers reaches 96.4%, confirming the practicality and effectiveness of this method.