Enhancement of coherency identification techniques for power system dynamic equivalents

In dynamic studies of large electric power systems, it is often necessary to reduce a large power network to a smaller equivalent system while keeping the dynamic characteristics of the full system model within the desired accuracy. This paper proposes new coherency identification techniques using the K-means algorithm for construction of dynamic equivalent models. Numerical simulation results based on a NPCC 48-machine test system are presented to demonstrate the performance of the proposed method.