Multi-source Heterogeneous Data Joint Diagnosis Method for Transformers Based on D-S Evidence Theory

This paper focuses on transformers and addresses their characteristics such as complexity, fuzziness, and uncertainty. We propose a multi-source heterogeneous data joint diagnosis method for transformers based on BP neural networks and the D-S evidence theory. Firstly, a BP neural network model is established for the preliminary diagnosis of transformer fault types. Subsequently, the diagnosis results from the BP neural network are fused using the D-S evidence theory. Discounting operation is applied to modify the fusion process in the evidence theory. Finally, the Basic Probability Assignment (BPA) is transformed into a probability distribution, with the highest probability indicating the identified fault.