The fault diagnosis of transformer based on BP neural network

The paper researches online assessment and fault diagnosis method of running transformer based on the BP neural network. The six types of gas content data: H₂, CH₄, C₂H₄, C₂H₂ and CO, is the input of BP neural network. There are seven kinds of failure: low-energy discharge, high-energy discharge, partial discharge low-temperature overheating, middle-temperature overheating, high-temperature overheating, high-temperature overheating and high-energy discharge. With 226 set of observational data on the neural network training, BP neural network model of running transformer´s online assessment and failure diagnosis can be obtained. The experimental results show that running transformer´s online assessment and failure diagnosis method based on BP neural network achieves a relatively high accuracy of failure diagnosis.