Distributed Optimisation for Energy Management in Building Networks

Abstract

We formulate and solve an energy management problem for a cooperative network of buildings sharing some resource and communicating over a time varying network, each of which desires to maintain its privacy. As the network is time-varying, we are able to accommodate communication constraints or failure between agents. We begin by introducing validated bilinear modelling techniques for individual buildings, and continue by showing how the energy management problem involving a network of buildings can be addressed using a distributed optimisation algorithm recently proposed in the literature, with individual buildings treated as agents. To facilitate these assumptions, we linearise the bilinear models generated, and introduce linear local and coupling constraints to model a shared thermal energy storage device. We formulate a linear objective function to minimise based on the cost of energy used by the agent not taken from the storage. Finally, we demonstrate the efficacy of the distributed algorithm as applied to the energy management problem using an extensive simulation based study.