From Islanding Detection to Islanding Identification: A Critical Step Toward Self-Healing Distribution Networks With Grid-Forming Inverter Fleets

As resilience becomes increasingly of concern, the traditional islanding detection concept for anti-islanding protection of Inverter-Based Resources (IBRs) becomes insufficient. When part of a distribution system forms an island, grid-forming IBRs could stay online to form a microgrid, maintaining the local frequency and voltage without a blackout stage. However, in order for this to happen, IBRs must identify the boundaries of islands, so they can establish microgrid secondary control with the right set of IBRs: those in the same island. In this paper, we propose the island identification concept as an significant extension of the traditional islanding detection concept. Upon disturbance detection, each IBR injects a small signal with its signature frequency into the grid, and tries to capture signature frequencies of other IBRs. As signals can be transmitted with each island, the sets of IBRs within the same island can be identified, and the boundaries of all islands can be determined. This concept is realized by an event-triggered, self-decaying, negative-sequence signal injection method to minimize the influence on grid power quality and IBR power control, as well as a highly effective Estimation of Signal Parameters via Rotational Invariance Technique (ESPRIT) for signal extraction. This framework is validated using on IEEE 13-node and 34-node test feeders with multiple IBRs and various islanding conditions.