Hybrid power solution modelling based on artificial intelligence

Power electronics become increasingly resourceful as the use of renewable energies increases. Microgrids and active distribution networks include various controllable devices that interact and may create instabilities. This underlines the necessity of modeling complex systems to conduct system-level analyses. As a first step toward tools for modeling inverter-based electrical systems, this paper introduces a model of the HyPoSol system, put in perspective with measurements on the real system. The HyPoSol system consists of a photovoltaic (PV) inverter, a battery, and a three-port converter designed by CE+T Power. To develop a model of the PV inverter, we employed an enhanced polytopic model which uses neural networks as weighting functions. The PV inverter model is combined with a Tremblay's battery model and a simplified model of the three-port converter. We conduct system-level analyses on the overall representation of the HyPoSol system and compare the results with measurements.