A Case Study on Clustering Methods Applied to Identification of Generator Coherency for Controlled Islanding

摘要

The controlled islanding is one of the essential tools that can be used to prevent a system-wide outage in a power system. Most of the controlled islanding strategies require generator coherency information in order to minimize the possibility of generators swinging against each other within an island. The generator coherency assessment is often considered as a separate problem because of its use cases such as model reduction, instability mitigation, and so on. The coherency assessment delivers different results depending on the partitioning method employed. A change in coherency assessment reflects in the usability of the corresponding controlled islanding solution. The objective of this paper is to study the usability of the controlled islanding solutions determined based on coherency information derived using different partitioning methods. In this paper, hierarchical and spectral clustering methods are applied to the coherency identification problem. Suitability of the clustering methods is studied on 68-bus and 300-bus test systems.