LAUSR

Current distribution system outage management is to separately use digital relays at the substation for estimation of fault location and use meter information to infer the activated protection device. The lack of holistic utilization of available data in the distribution operating center results in lost opportunities for accurate fault diagnosis. To solve this issue, this study proposes a data-driven probabilistic fault location methodology based on comprehensive sensing measurement from digital relays at substations, Intelligent Electric Devices (IEDs) along primary feeders, SCADA sensors in the feeder circuit, and smart meters at customers’ premises. Statistics of historical fault location accuracies by digital relays and IEDs are used to estimate fault location errors with probability in real time. Multiple-hypothesis analysis is implemented to handle the uncertainties from SCADA sensors and smart meters. The spatial correlation between the potential fault location and collected sensor data is modeled as a mixed integer linear programming (MILP) problem. By solving the proposed optimization for generated hypotheses, a list of potential fault locations with possibilities is provided to system operators in decision-making for facilitated fault isolation and service restoration. Simulation results with a utility feeder validate the efficacy of the proposed approach for fault diagnosis.