Voltage Sag Diagnosis Using Big Data Analysis

We proposes a data-driven approach for voltage sag diagnosis in this paper. Rather than traditional features such as frequency, amplitude and duration, we apply temporal distribution as a new feature to distinguish whether the sag is caused by power system faults or heavy load switching. Sags caused by these two reasons have different distribution pattern, and this work interprets it from a number of perspectives. We also perform voltage sag source location by clustering. This approach don't use any physical level analysis and can find the fault source faster as well as accurately.