Fault diagnosis of oil‐immersed transformers based on the improved sparrow search algorithm optimised support vector machine

Abstract

Oil-immersed transformers play an important role in the stable operation of power systems. Aiming at the low accuracy of traditional transformer fault diagnosis methods, a transformer fault diagnosis method using an improved sparrow search algorithm (ISSA) optimised support vector machine (SVM) is proposed. First, use Sin chaotic to initialise the population, then introduce the fusion of Cauchy mutation and opposition-based learning to optimise the selection of the population to improve the SSA global optimisation capability. Secondly, use the improved Sparrow algorithm to optimise the SVM kernel function parameters and penalty coefficients and establish a fault diagnosis model--ISSA-SVM for dissolved gas analysis. Enter the data into ISSA-SVM for fault diagnosis and combine it with K nearest neighbour (KNN) algorithm, gradient boosting decision tree (GBDT), sparrow search optimised deep extreme learning machine (SSA-DELM), SVM, sparrow search algorithm optimised Support vector machine (SSA-SVM), and other diagnostic results are compared. The results show that the fault diagnosis rate of ISSA-SVM is 91.43%, which is 12.86%, 7.14%, 5.71%, 2.86%, and 1.43% higher than that of KNN, GBDT, SSA-DELM, SVM, and SSA-SVM, respectively. It can more accurately judge the current operating state of the transformer.