 |
Center for Nonlinear Studies |
In this talk we describe ongoing work with new methodologies for two classes of problems: (1) vulnerability analysis of large-scale transmission systems, and (2) algorithms for throughput maximization in transmission systems. Vulnerability analysis, in particular the so-called "N-k" problem and derivatives, are well-known. As mathematical problems these are quite difficult. We first present results with a mixed-integer programming formulation that addresses a standard version of the problem. We then present results with an indirect approach that proves vastly more scalable and informative while at the same time being able to address a more realistic version of the problem, including 'noise' and model uncertainty. Concerning (2) we describe a primal-dual approach which combines techniques from convex programming, linear programming and branching techniques.
Host: Misha Chertkov