Study on Transformer Fault Diagnosis Based on IAVOA

Based on the low accuracy of transformer fault diagnosis and less complexity, a research algorithm based on IAVOA is proposed. Firstly, obtain relevant data through gas decomposition in oil (DGA) and form 24-dimensional features, integrate the core principal component analysis to solve the problem of the compressor. Secondly, improve the African Vulture optimization algorithm (IAVOA), improve its optimization ability, avoid the local optimization, to optimize the parameters and weights of the kernel function in the hybrid core limit learning machine (HKELM) to increase the classification accuracy and improve the generalization ability and accuracy. The test results showed that the IAVOA-HKELM model achieved 96% accuracy, which improved by 10% over the ABC-HKELM model, 8% and 7%, respectively.