Sep 2014


Our paper “Hierarchical Spectral Clustering of Power Grids” has now appeared in IEEE Transactions in Power Systems:

The paper uses spectral methods to decompose power networks into reasonably self contained units, with the aim of protecting the grid during (or preventing) cascading blackouts. The method is fairly general and gives a fast efficient algorithm to iteratively decompose a weighted network into pieces which maintain balanced flows. In the blackouts application the network is weighted by power flow, typically using a DC approximation, to ensure that the resulting components are self sustaining, but other applications may well be possible.