Optimization of KELM for Transformer Fault Diagnosis Based on Dung Beetle Optimizer Algorithm

Aiming at the problem of inaccurate classification accuracy of existing fault classification models, a Dung Beetle Optimizer (DBO) method is proposed to optimize the Kernel Extreme Learning Machine (KELM) for transformer pattern recognition. Firstly, to address the issue of the KELM method being greatly affected by parameter values during fault classification. The kernel function and regularization coefficient of KELM are determined by using DBO to establish a DBO-KELM diagnosis model, and the fault diagnosis is carried out by using the optimized DBO-KELM diagnosis model, and compared with ABC-KELM and GWO-KELM. The final experimental results show that the diagnosis effect of DBO-KELM is better and the accuracy is higher.