Algorithmes asynchrones pour la gestion décentralisée des réseaux électriques soumis aux aléas de communication

Smart grids are based on the interaction of an electrical network and a communication network between distributed, "intelligent" and communicating agents. In the case of a traditional electrical network, unforeseen events (loss of an equipment, line congestion, etc.) are mainly absorbed and compensated by the inertia of the network and its operating reserve. Smart grids aim to reduce this dependency by using intelligent distributed management strategies using information exchange and therefore based on a communication network. The impact of the communication network's reliability is a critical issue that we must address. This project aims at analyzing the impact of communication hazards on the resilience of a smart grid and at developing strategies that take these hazards into account in order to guarantee the operability and efficiency of the smart grid. In particular, we investigate three types of algorithms: a peer-to-peer market algorithm, an Optimal Power Flow algorithm that takes into account the physical constraints of the power grid, and finally an endogenous peer-to-peer market algorithm. We have studied in depth the principles of asynchronous algorithms, which help to minimize the waiting time at each iteration, as well as the effects of communication network hazards on the asynchronous resolution of these algorithms.