应用于油中溶解气体分析的深度信念网络与典型神经网络对比研究

Dissolved gas analysis in transformer oil is one of the most effective methods to diagnose the fault of oil-immersed power transformer insulation system, which is of great significance to realize the accurate diagnosis of transformer fault. Based on the deep belief network(DBN) and the typical neural network model, BP neural network(BPNN)and RBF neural network(RBFNN), this paper builds different structures of the DGA models to realize the transformer fault diagnosis. In addition, influences of different training data, modeling methods and parameters on the diagnosis accuracy of transformers is further analyzed. The proposed dissolved gas in oil model based on DBN achieves the highest recognition accuracy of 84.87% under 1 000-set training data, 5-layer in DBN model structure and 20 nodes in 2nd-layer.