Power big data mining with local technique: a survey on power faults

The power grid is a fundamental infrastucture for an urban system, but is usually challenged by frequent power faults. In this study, we adopted the power grid big data and a localized statistical approach, namely geographically weighted regression (GWR) to examine the spatially heterogeneous relationships between power fault (PF) and its influencing variables with the population and point of interest (POI) data. To differentiate the temporal effects, we conducted the GWR models with the monthly PF data. Results show that the PF could be affected with both spatially and temporally heterogeneous characteristics.