Hydrogen backup and peak-shaving: Multi-energy modular architecture with AI and digital twin supervision

This paper analyses the performance of an artificial intelligence-based energy management system, specifically designed to optimize the operation of hybrid energy backup systems integrating fuel cells and batteries. Using customized genetic algorithms, the energy management system calculates the optimal operation setpoints for minimizing energy costs in a consumption section of a hospital environment subject to time-dependent electricity prices. The energy management system is designed to (1) support critical loads during power outages, (2) manage peak power demand to allow for a reduction in the contracted maximum capacity, and (3) to shift part of the intraday electrical load to low-price hours. The system considered includes an electrolyzer, fuel cells, a hydrogen tank and electric batteries.