Transformer Fault Diagnosis Method Based on Deep Belief Networks and DSmT
Haochuan Fu, Qimeng Liu, Yuqiang Wang, Zheng Gao, Jia Liu, Chong Wang, Wei Guo Song, Yonglin Li · 2023
According to the deficiency of traditional machine learning theory and the present situation that the state features of reference are single in transformer fault diagnosis, a transformer fault diagnosis method is put foward based on Deep Belief Networks(DBN) and DSmT (Dezert-Smarandache theory). The on-line monitoring data and test data which can reflect the transformer fault information are chosen as the diagnostic parameters. A parallel training unit is constructed with DBN and DSmT (Dezert-Smarandache theory). training unit is constructed with DBN to construct the basic belief assignment (BBA) for the transformer fault recognition framework. Based on the idea of information fusion, we can get the basic belief assignment for the transformer fault recognition framework. Based on the idea of information fusion, we can get the final diagnosis conclusion applying DSmT theory to fusing BBA with the family defect record, which overcomes the limitations of D-S evidence theory that the family defect record can be identified by the BBA. limitations of D-S evidence theory that can not solve the fusion problem of high conflict evidence. Through an example of a 110kV transformer, the result shows that this method has good practicability.