Distributed optimisation with restricted exchanges of information: charging of an electric vehicle fleet

Abstract

The multiplication of flexible loads into electrical power systems allows to develop distributed optimisation techniques. The goal is to achieve the best compromise between the individual objectives of each load and the general interest which could be for instance the safety of the grid or the efficiency of the power generation. However the algorithms of distributed optimisation mostly rely on an iterated exchange of information, such as the dual variable and the decision of each agent. This exchange therefore supposes a bidirectional communication which cannot be considered as granted in operational contexts. This paper investigates the utilisation of the Alternating Direction Method of Multipliers under restricted exchanges of information. The case study consists in an electric vehicle fleet which has to comply with a maximum power limit. It is established that a close to optimal behaviour can be achieved using frequent broadcast messages instead of duplex communication. Keywords: Electric vehicle fleet, Alternating Direction Method of Multipliers, Stochastic Dynamic Programming.