A novel fault diagnosis method for transformer
Ge Hui · Light Metals · 2010
A novel reduction method is put forward based on improved rough set to deal with the difficulty of on-line fault diagnosis on transformer. The value and ratio of dissolved gas-in-oil is seemed as original rules for the final results. The problems of continuous attribute discretization, attributes reduction and attribute values reduction of rough sets are translated into the only one thing of discernibility matrix simplified. In this way, the precision, efficiency and consistency of fault identification can be improved. The schematic diagram of fault diagnosis on transformer is presented. The satisfying fault diagnosis results can be obtained by the experiment on the 10kV main transformer (dissepiment), which shows that the method in the paper is effective and practical.