Stator winding's inter-turn fault intelligent diagnosis in large turbo- generator by Elman neural network

Abstract

Turbo-generator stator's inter-turn short is a usual serious fault, there would have hidden big trouble for electric power system's safety due to lack of efficient protection. On-line monitoring generator's operate condition combined intelligence non-line identify technology is presented to observe fault in time instead of poor function of protection. Longitudinal zero-sequence voltage and fault phase's current are analysis as stator winding's inter-turn short's stable fault characters, mathematical model of which are build, Elman neural network which do well for dynamic data in real time are introduced to identify the fault. A large turbo-generator's general parameters are used for calculate its stable fault characters during stator winding's inter-turn short occur in operation, and identification are performed by trained Elman neural network followed. Example indicate that the Elman network could efficiently identify generator stator's inter-turn short based on rational fault characters combine.