Transformer Top-oil Temperature Modeling Based on Kernel-based Extreme Learning Machine

Abstract

Transformer top-oil temperature (TOT) and winding hot-spot temperature (HST) are key indices to evaluate thermal condition of transformers. In order to improve TOT prediction accuracy, a TOT prediction model based on kernel-based extreme learning machine is established and particle swarm optimization algorithm is adopted to train the model and optimize the kernel parameters. The proposed model is tested on a 50MVA 110/37kV ONAN transformer. Besides, to verify the advantages of the proposed model, it’s compared with several traditional data-driven models. The results demonstrate the validity and accuracy of the proposed model.