Abstract: The expert system used in power system fault diagnosis has a long development cycle and can not diagnose the faults beyond experiences. A kind of fault diagnosis of power system was proposed based on model-based diagnosis and skyline query. The diagnosed system was decomposed into several independent subsystems based on the distribution of measurement points. By establishing system models considering the causality between the system failure output and system element， candidate diagnosis of each subsystem was inferred. The actual alarm information was introduced into diagnosis logic framework to identify the fault element from candidate diagnosis by using skyline query. Ready candidate diagnosis was obtained offline and candidate diagnostic was confirmed online based on the proposed method， which saved the time of diagnosis. Actual alarm information was introduced into diagnosis logic framework， which improved the efficiency of diagnosis. The simulation results show that the method can effectively reduce the complexity of time and space.

Key words: power system fault diagnosis, model-based diagnosis, candidate diagnosis, alarm information, skyline query