This PhD research investigates the optimisation of Energy Management and Optimisation Systems (EMOSs), i.e., automated systems that coordinate the optimal operation of various on-site energy sources, with the goal of reducing energy costs. The increasing deployment of renewable energy sources is widely recognised as a crucial element in mitigating the impacts of climate change. Nevertheless, their variable nature, driven by weather dependency, introduces significant challenges for energy markets and system operations. Our focus is on improving forecasts of photovoltaic (PV) production and electricity consumption, which serve as key inputs for energy optimisation. Advanced data-driven techniques, including deep learning, will be explored to improve the accuracy of consumption and grid injection predictions. These enhanced forecasts will inform optimisation strategies that enable the intelligent coordination of flexible assets such as batteries, electric vehicles (EVs) and heat pumps. The economic and environmental benefits of such optimisation strongly depend on the degree of user flexibility, since fewer constraints allow significantly higher gains. This research will investigate strategies to improve these optimisation models leveraging real word data. The goal is to improve energy consumption planning, reducing energy costs, maximising the utilisation of renewable energy, and thereby contributing to methodological advancements and practical solutions for sustainable energy systems.
