Minimal Controlled Islanding With Similarity-Based Coherency Identification Using Phasor Measurement Data

Controlled islanding is the last protective scheme to prevent passive islanding or total system blackout caused by cascading failures. In this article, a novel coherency detection approach is proposed, which is based on a density and distance clustering algorithm. The coherency detection approach is parameter free and requires only the frequency measurement data since the approach utilizes the cosine similarity combined by the Gaussian principle. Following the coherency detection, the proposed island reduction algorithm utilizes frequency deviation at center-of-inertia reference and locates the fault-affected generators plus the faulted area to construct the islanding scenarios. To find a practical splitting strategy with better network reconnection, an island reduction algorithm is presented to reduce the number of islands and isolate the faulted area from the rest of the system considering the coherency and system operational constraints. Additionally, an mixed integer liner programming (MILP) optimization model based on the linearized ac power flow is employed to find the optimal island boundaries and guarantee the efficient isolation of critical or most affected islands. The efficiency and performance of the proposed islanding approach are tested on the IEEE 118-bus test system.

Automated summary

This article proposes a novel coherency detection approach based on a density and distance clustering algorithm, as well as an island reduction algorithm. By utilizing this approach for coherency detection and island reduction, passive islanding and system blackout can be effectively prevented, and the efficiency of network reconnection can be improved.