Research on Fault Diagnosis of Neural Network Power Transformer Based on Dung Beetle Optimization Algorithm

In this study, a fault diagnosis method for power transformers based on dissolved gas analysis (DGA) was proposed. Firstly, nuclear principal component analysis (KPCA) is used to preprocess the collected fault data to remove the interference data, and KPCA is used to perform feature extraction on the mixed DGA data. Then, the dung beetle optimization algorithm (DBO) was used to optimize the neural network algorithm (BP), and an improved dung beetle optimization algorithm (DBOBP) was formed to achieve better optimization accuracy and convergence speed. Since tent diagrams are used instead of traditional population initialization methods, this method improves population diversity. Simulation examples verify the superior performance of the proposed method, including high diagnostic accuracy, short diagnosis time, strong significance and effectiveness. This study provides a feasible research idea for solving practical engineering problems in the field of power transformer fault diagnosis.