Turn-to-Turn Fault Localization of Power Transformers Using Neural Network Techniques

Power transformers are considered as one of the essential elements in electrical networks. Any failure in these equipments directly reduces network reliability and increases maintenance costs. Therefore, quick and precise diagnosis of faults is very important. One of the main faults in this regards is turn-to-turn fault that its localization is very useful and vital. This paper has been presented a novel method for diagnosing and localizing of above mentioned faults in power transformers using both transfer function method and neural network technique. In this contribution some different turn-to-turn faults have been simulated on a laboratory HV winding of power transformer and correspond their transfer functions have been measured. Then, using an ANN with Back-Propagation learning algorithm, a method has been proposed to identify the local of occurred fault. Results show that the ANN could well diagnose the local of fault if it is trained using the presented patterns.