Groupe d’études et de recherche en analyse des décisions

G-2018-39

Coupling decomposition with dynamic programming with application to an energy model

, et

This paper targets a stochastic energy management problem. We first decouple the stochasticity of the global scenarios to local scenarios. Then, we use spatial decomposition through ADMM and time decomposition through Dynamic Programming to find approximate optimal values. The algorithms are validated on toy problems and a convergence discussion about the coupling of ADMM and Dynamic Programming is started.

, 20 pages