Research On Operation Fault Diagnosis Models For Transformers

Transformer power system as an important equipment. It is safe, reliable and stable operation of the power system has an important role, it is possible to early prediction and detection to the transformer potential fault to improve the safe operation of the power system has important significance.Dissolved gas analysis(DGA) is a method of fault diagnosis of large oil immersed transformer is used. According to the characteristics of gas and fault symptom of deep connection, according to the fault generated content of characteristic gases for the establishment of the transformer fault, but due to the transformer fault often crisscross and fault symptoms and characteristics of gas between uncertainty relation, so the traditional fault diagnosis methods, such as: three ratio method, improved three ratio method, ratio method and David triangle method are its defects. The modern intelligent algorithm can remedy this problem, as the nonlinear relationship between the input and output of the BP neural network can be used widely.This paper in transformer oil dissolved gas technology based on and under the MATLAB environment using neural network toolbox to establish the model of transformer fault diagnosis based on BP neural network, many simulation experiments to determine the hidden layer neuron number, and ultimately determine the BP God 5-11-3 structure of the network diagnosis model. The training error of the LM is reduced to 0.1, and the accuracy of the corresponding diagnostic model is 90%, and has a certain practical value.Established based on BP neural network model of transformer fault diagnosis, this article uses the Matlab GUI design a graphical user interface window, transformer fault diagnosis algorithm embedded in them, the interface can guide users to adopt different methods for fault diagnosis, using modified good three ratio method and BP neural network and the added bonus of background, alarm, help other additional features. And according to the user’s preferences, the fault diagnosis report can be generated for the user to handle the transformer status data in batches.