Improved Genetic Algorithm and XGBoost Classifier for Power Transformer Fault Diagnosis

Power transformer is an essential component for the stable and reliable operation of electrical power grid. The traditional fault diagnostic methods based on dissolved gas analysis are limited due to low accuracy identification. In this study, effective diagnosis system proposed improve identification accuracy. approach combines improved genetic algorithm (IGA) with XGBoost form a hybrid network. combination (IGA-XGBoost) forms basic unit method, which decomposes reconstructs recognition problem into several minor problems IGA-XGBoosts can solve. results simulation experiments show that IGA performs excellently in combined optimization input feature selection parameter, method accurately identify types average 99.2%. Compared IEC ratios, dual triangle, support vector machine common by 30.2, 47.2, 11.2, 3.6%, respectively. be potential solution types.