Research on Transformer Fault Diagnosis Based on Improved INGO Optimization LSSVM

In order to improve the accuracy of transformer fault diagnosis, an improved Northern Goshawk optimization algorithm (INGO) is proposed to optimize the least squares support vector machine (LSSVM) model, and reduce the influence of invalid features by dimensionality reduction of acquired transformer fault data through kernel principal component analysis. The fault data processed by KPCA was input into INGO-LSSVM diagnosis model for fault diagnosis simulation research. The simulation results showed that the accuracy rate of INGO-LSSVM model reached 94.17%, which was 7.5% and 11.67% higher than that of NGO-LSSVM and PSO-LSSVM models. It is proved that the proposed method can effectively improve the performance of transformer diagnosis.