Enhancing Fault Diagnosis in Transformers by Combining HHO and RF

In order to solve the problem of low accuracy of existing artificial intelligence algorithms in judging the type of transformer faults, this paper proposes a method based on Harris Eagle (HHO) optimization Random Forest (RF) to classify and predict transformer faults. Firstly, the non-coding ratio method was used to construct a new data set for the fault gas data, and the characteristic parameters were normalized, and the training set obtained after division was used as the input of the model. Then, the two hyperparameters in RF were iteratively optimized by HHO to obtain the optimal parameter values of the diagnostic model. Finally, the diagnostic effects of the RF model with the optimal parameters were compared with those of NB, KNN and RF models under different feature parameter inputs. Experimental results show that the HHO-RF method has the best effect on transformer fault diagnosis, and the generalization ability of RF model after optimizing parameters is improved, and the diagnosis accuracy is improved.