Data-driven online distributed disturbance location for large-scale power grids

Timely detecting disturbances and locating their sources are critical to the reliable operation of power grids. This capability enables operators to effectively diagnose disturbances over wide areas and earns time for remedial reactions. In this study, a travelling-wave based scheme, namely data-driven online distributed disturbance location (DODDL), is proposed to quickly detect disturbances and determine their geographic location in large-scale power grids when the grids' topology is not available. The proposed DODDL scheme consists of two function blocks: (i) a singular spectrum analysis-based change-point detection method, which can quickly detect disturbances and determine their arrival time at distributed sensors, and (ii) a novel temporal scanning algorithm, which can accurately determine the geographic location of the disturbance source point. Utilising field measurement data sets recorded by the frequency disturbance recorders from the frequency monitoring network, it is shown that the DODDL scheme is not only quicker and more robust to grid non-homogeneity than existing approaches, but also can capture and locate more subtle and concealed disturbances.