Spectral clustering of power transmission networks Seminar

A network can be used to represent a complex system as a pattern of interactions between components. The eigenvalues of this network (mathematically, a graph) can be used to extract useful global information of the underlying complex system. For example, the eigenvalues of the graph laplacian matrix can reveal clustering structure and thus help to identify a hierarchy of subsystems.

We present some recent advances on the theory of spectral clustering and explain how it can be adapted to the study of power transmission networks. The underlying motivation is to prevent cascading failures leading to wide-area blackouts. This is joint work between the universities of Southampton (with Jacek Brodzki, Graham Niblo, Nick Wright and Max Fennelly) and Durham (Janusz Bialek, Sean Norris).