Power transformer is a very complicated grey system

and the relevance between fault symptoms and cause is complex. Therefore

fault diagnosis of transformers is difficult and of low accuracy. In this paper

a novel model based on weighted degree of grey incidence of optimized entropy is put forward for the diagnosis. The improved grey incidence has the advantages of improving its reliability and accuracy

and reducing the error brought by system interference. Experiments are conducted to test the effectiveness of the proposed model using hydrogen

methane

acetylene

ethane

and ethylene gases decomposed from transformer oil. Diagnosis results demonstrate that the new model can find the type of fault with higher accuracy than the three-ratio method and other methods based on grey incidence.