Contributions à la gestion des réseaux de distribution intégrant massivement des flexibilités distribuées

With the rise of renewable energy and distributed flexibility (DF), this thesis examines how transmission network (TN) management strategies can be adapted to future distribution networks (DN). Historically passive and unconstrained, DRs will need to find new levers in the future to maintain the quality of energy supply. It is therefore crucial to rethink power flow management, not only to continue to respect voltage and thermal constraints, but also to fully exploit the potential of DF. The manuscript is structured around three key questions : How do impedance uncertainties associated with conductor ageing affect DN management? How can robust management strategies be developed without detailed knowledge of the behavior of network actors? How much information is needed to efficiently manage a distribution feeder ?More specifically, the proposed solutions are based on optimal power flow and machine learning techniques. Emphasis is placed on the operability of the control methods, in particular by reducing the number of variables to be managed through dimension reduction techniques.