Abstract:  To overcome the mis-diagnosis of fault caused by the distortion of rea- l time diagnosis information during the generation and transfer processes‚on the basis of the research into the RST （Rough Set Theory）-based fault diagnosis model of high voltage transmission line system and by utilizing the generalization ability of neural network （NN） and the great qualitative analysis ability of RST‚the fault diagnosis model with the combination of RST and NN was constructed．In this approach‚RST was used to extract knowledge region data set from diagnosis samples‚and the basic structure of NN was built on the basis of diagnosis knowledge attribute．Thus the intelligence and fault tolerance performance of diagnosis NN system were improved．The validity and commonality of the proposed model were proved by the comparison of simulation results of the fault diagnosis system．The proposed model has excellent fault tolerance performance even though the diagnosis information is not complete‚so it is of important practical value in the rea- l time fault diagnosis of electric power system．

Key words:  transmission line system, fault diagnosis, fault tolerance performance, rough set theory, neural network