Abstract:

Based on the oil gas feature content, this paper contracts the BP neural network and acquires the data sample to train the neural network. Then a layered fault diagnosis model is  established with the information fusion theory. This method can diagnose the transformer fault type and recognize the discharge type. This paper develops a fault diagnosis system for the transformer based on the virtual instrument technology. The system diagnoses transformer faults and recognize partial discharge types from gas quantity and partial discharge signals. To diagnose transformer faults by oil gas content and at the same time use a statistical method to extract the twodimensional operator, we apply information fusion to judge the transformer partial discharge type. The system realizes the transformer oil and gas content analysis, partial discharge signals storage, data processing, database query, modification and deletion function, thus fulfills the transformer fault diagnosis.

Key words: transformer;fault diagnosis;partial discharge;oil gas content;Labview