Edge/cloud virtualization techniques and resources allocation algorithms for IoT-based smart energy applications.

Nowadays, the installation of residential battery energy storage (BES) has increased as a consequence of the decrease in the cost of batteries. The coupling of small-scale energy generation (residential PV) and residential BES promotes the integration of microgrids (MG), i.e., clusters of local energy sources, energy storages, and customers which are represented as a single controllable entity. The operations between multiple grid-connected MGs and the distribution network can be coordinated by controlling the power exchange; however, in order to achieve this level of coordination, a control and communication MG interface should be developed as an add-on DMS (Distribution Management System) functionality to integrate the MG energy scheduling with the network optimal power flow. This thesis proposes an edge-cloud architecture that is able to integrate the microgrid energy scheduling method with the grid constrained power flow, as well as providing tools for controlling and monitoring edge devices. As a specific case study, we consider the problem of determining the energy scheduling (amount extracted/stored from/in batteries) for each prosumer in a microgrid with a certain global objective (e.g. to make a few energy exchanges as possible with the main grid). The results show that, in order to have better optimization of the BES scheduling, it is necessary to evaluate the composition of a microgrid in such a way as to have balanced deficits and surpluses, which can be performed with Machine Learning (ML) techniques based on past production and consumption data for each prosumer.