Intelligent system applied in diagnosis of transformer oil

The accurate diagnosis of transformer oil is very important to establish the degree of the aging of transformers. Several experimental tests and theoretical analyses have been carried out to obtain parameters associated with the advances on understanding failure processes and regeneration systems. This paper describes a novel approach for mapping diagnosis of oil using an intelligent system based on artificial neural networks. The network acts as an identifier of structural features of the oil so that output parameters can be estimated and generalized from an input parameter set. This set takes into account several factors, such as interfacial tension, density, oil temperature, humidity, pressure, furfural level and so on. The results obtained by the network are compared with those which had been provided by tests of chromatography in the laboratory.