Transformer Fault Diagnosis Method via Approximation Relations in Approximation Space
Feng-Bo Tao, Tonglei Wang, Yao-Yu Xu, Chao Wei, Yuan Li, Guanjun Zhang · 2019
Although many techniques now are available for transformer fault diagnosis, one of the main issues need to be further investigated, i.e. how to address the incomplete and uncertain monitoring information in a fault diagnostic task. In this paper, we propose a transformer fault diagnosis method via approximation relations in approximation space to accomplish decision-making under incomplete information. Firstly, we build a decision-making table of transformers based on Rough Set (RS) theory in which each decision-making rule includes some conditional attributes and a correspondingly decision attributes. Hence, approximation relations are used to calculate the dependency of attributes in the approximation space, which provide the criterions to determine the optimal reduction sets of the table. When the conditional attributes in a diagnostic task are determined by monitoring information, we can use the reduction sets to match the task for obtaining the diagnostic results. It comes to conclusion that this proposed method shows a promising results of transformer fault diagnosis with high accuracy of 75.41% under incomplete information. In addition, the method could be improved by new symptoms-fault knowledge discovered.