Fault diagnosis using hybrid artificial intelligent methods

This paper presents genetic-based neural networks (GNNs) for fault diagnosis of power transformers. The GNNs automatically tune the network parameters, connection weights and bias terms of the neural networks, to yield the best model according to the proposed genetic algorithm. Due to the global search capabilities of the genetic algorithm and the highly nonlinear mapping nature of the neural networks, the GNNs can identify complicated relationships among dissolved gas contents in transformer oil and corresponding fault types. The proposed GNNs have been tested on the Taipower Company diagnostic records and compared with the fuzzy logic diagnosis system, artificial neural networks and the conventional method. The test results show that the proposed GNNs improve the diagnosis accuracy and the learning speed of the existing approaches.