Dynamic state and input aggregation

In this paper we study a model reduction technique for sparse networked energy systems. Our method differs from standard model reduction techniques in that it aims to preserve the sparsity of the system and also preserves the total energy of the system. We apply our technique to constrained and soft-constrained linear optimal control problems. We show through numerical examples that our method is comparable to standard model reduction techniques in terms of sub-optimality but have faster solution times because of its ability to preserve sparsity.