Power system faults with the possibility of threatening public safety are of highest importance. High impedance faults (HIFs), due to the companionship with electrical arcs, occur on power distribution system feeders with close proximity to the population. The randomness, low current, non-linearity and dependency towards network topology have commonly resulted in failed detection. Moreover, typical system events such as switching events have shown the same pattern which has led to false detection and major distrust in available detection schemes. In this paper, a new HIF detection algorithm based on the concept of Evidential Reasoning (ER) has been proposed. The employment of various basic attributes with different advantages results in a wider detection scope and higher detection rates. The designed scheme has the ability to successfully detect and distinguish switching events from HIFs. Rigorous simulations via the IEEE 13-Node standard distribution system and experimental real world HIF data acquired from southwest Tehran Distribution Company have provided proof to these claims. Simulation trails in low current single phase extensions have also shown promising results. Unaffectedness in regards to network topology, fault resistance values, fault position and lack of sufficient information regarding basic attributes are additional virtues of the proposed method.

Keywords