Disturbance detection in the phase space through GK clustering

Power disturbances have always been an important issue in power system protection. This paper proposes an original approach to detect the disturbances in power system signals. The disturbance signal is firstly transformed to the phase space, where the disturbance-free signal and the disturbance form two irrelevant waveforms. GK clustering is used afterwards to distinguish the two clusters. Simulation studies are conducted on a batch of power system disturbances, which show that the proposed approach is capable of detecting and localizing the various disturbances of power systems.