A new kernel-based clustering algorithm for the multi-machine equivalent of large power systems

Formulating the power system equations with respect to the systempsilas inertial center, it is shown that a large disturbance tends to pull each generator from the rest of the generators, forming a multi-machine equivalent. Clustering generators is a useful general technique in data mining of large-scale power grid data sets. This paper presents a new kernel-based clustering algorithm for multi-machine equivalent of large power systems. Based on the multi-machine equivalent concept, transient stability assessment of a multi-machine power system is made, and a criterion for identifying the generators likely to separate from the system (by loss of synchronism) and the energy associated with this separation is developed. The criterion is applied successfully to a 491-generator power system. Experimental results indicate that the new algorithm is efficient and effective at finding both good clustering and the appropriate number of clusters.