Distributed MDP for water resources planning and management in inland waterways

International audienceInland waterway management should undergo heavy changes due to a commitment to increase the waterway traffic in a context of climate change. These new constraints will impose an adaptive and resilient management of the water resource. The aim is to plan optimally the water resource distribution over the integrity of the inland waterway network, while taking into account the uncertainties arising from their operation. Due to the large size of waterways, a centralized modeling would not be able to represent an entire network. A distributed Markov Decision Process modeling of inland waterways associated with a resolution algorithm is proposed to allow full scalability, at the cost of optimality. The proposed approach is tested on a subnetwork composed of 7 reaches