Microgrid energy management systems design by computational intelligence techniques

With the capillary spread of multi-energy systems such as microgrids, nanogrids, smart homes and hybrid electric vehicles, the design of a suitable Energy Management System (EMS) able to schedule the local energy flows in real time has a key role for the development of Renewable Energy Sources (RESs) and for reducing pollutant emissions. In the literature, most EMSs proposed are based on the implementation of energy systems prediction which enable to run a specific optimization algorithm. Such strategy, known as Rolling Time Horizon (RTH), demonstrated very effective when the supporting prediction system performs well. However, it is featured by high operational times. In this work, different lightweight EMS models synthesized through machine learning algorithms have been compared considering six different simulation scenarios. Results shows that an RTH-based EMS owns the best overall performances. However, in some case studies, also other EMSs show competitive results, especially those based on Adaptive Neuro Fuzzy Inference Systems (ANFIS) trained by clustering, which in one case outperform RTH EMSs, and in other 3 cases (out of 6) yields performances close to RTH EMSs within 5%. A second contribution concerns the RTH EMS implementation on a small micro-controller, highlighting the high computational effort which can range in the order of minutes. Conversely, the ANFIS EMS shows always almost negligible computational costs (less than one second) and therefore can be used in realistic scenarios on cheap devices at run time. The paper also proposed a novel graphic tool to better represent, observe and analyze microgrid energy flows in each time slot or along the overall considered dataset.