Mining Transformer Fault Diagnosis Based on INGO Optimized LSSVM

In order to improve the diagnostic accuracy of mining transformers, an improved Northern Eagle Algorithm (INGO) optimized least squares support vector machine (LSSVM) fault diagnosis model for mining transformers was proposed. Firstly, KPCA is used to reduce the dimension of the obtained transformer fault data to reduce the impact of invalid features; Then, using the strategy of Logistic chaotic mapping, Cauchy mutation, and random difference perturbation, the traditional Northern Hawk algorithm is improved to improve its optimization ability. Compared with NGO and particle swarm optimization, it is proved that its convergence speed and solution accuracy have been improved; Finally, INGO is used to optimize the relevant hyperparameters in the solution set to improve the classification accuracy of LSSVM. The fault data processed by KPCA is input into the INGO-LSSVM diagnostic model for fault diagnosis simulation research. The simulation results show that the accuracy of the INGO-LSSVM model reaches 93.26%, which is 7.4% and 10.54% higher than the NGO-LSSVM and PSO-LSSVM models, demonstrating that the method in this paper can effectively improve the transformer diagnosis performance.