A Real-Time Combined Dynamic Control Framework for Multi-Energy Microgrids Coupling Hydrogen, Electricity, Heating and Cooling Systems

Abstract

While the energy management and control techniques have been extensively studied in electrical microgrids, optimizing electrical networks alongside other energy vectors, such as hydrogen, heating and cooling systems, remains a significant challenge. Effective real-time control management within multi-energy microgrids (MEMGs) is particularly challenging due to the intermittent and unpredictable nature of renewable energy sources and varying multi-energy demand. Existing research on MEMGs often lacks a holistic, real-time approach that simultaneously incorporates multiple intelligent techniques. Furthermore, the integration of co-generation systems, particularly those involving hydrogen and gas technologies, presents additional challenges in optimizing MEMG operations. This paper proposes a novel dynamic control strategy that directly addresses these challenges by integrating fuzzy logic (FL), model predictive control (MPC), and nonlinear optimization in real time. The strategy is designed to enhance MEMG performance by seamlessly coordinating multiple energy vectors, with a particular focus on the effective management of hydrogen storage and electrical batteries within a hybrid energy storage system (HESS). The objective is to minimize operational costs, gas consumption, and grid dependence, while maximizing system flexibility. The strategy is applied to an 8-unit residential building in Cardiff, UK, equipped with a photovoltaic plant, fuel cell, electrolyzer, hydrogen storage, battery, gas and electric boilers, chiller, and a combined heat-and-power unit. When compared to two alternative strategies—one that does not consider optimal cost allocation and another using a states-based EMS—the proposed framework yields a substantial reduction in costs by 33.86% and 18.38%. Gas consumption is reduced by 7.41% and 3.15%, respectively, while the HESS state of energy increases significantly by 100.06% and 20.02%, respectively. Furthermore, real-time experimental validation corroborates the practicality and efficacy of the proposed framework