A Novel Method Based on Particle Swarm Optimization Support Vector Neural Network for Transformer Fault Diagnosis

In order to solve the issue of low accuracy in transformer fault diagnosis, a novel method based on particle swarm optimization support vector neural network (PSO-SVNN) is proposed in this paper. Firstly, the transformer fault classification problem is constructed using support vector machine and transformed into a standard convex quadratic programming mathematical model. Then, a varying parameter recurrent neural network solver is employed for model solution. Finally, the particle swarm optimization algorithm is applied to iteratively search for the optimal penalty term (C) and kernel parameter ( \(\sigma \) ) in the model, aiming to improve the accuracy of transformer fault classification. Experimental results on IEC TC 10 dataset demonstrate that the proposed method outperforms traditional methods, achieving a classification accuracy of 86.3% with 5-fold cross-validation.