Title : Smart predictive control for sustainable and resilient hybrid microgrids
Abstract:
The decarbonisation of local energy systems requires not only the deployment of renewable energy sources, but also advanced energy management strategies that ensure their optimal, reliable, and sustainable operation. In this context, hybrid microgrids combining photovoltaic generation with distributed storage technologies emerge as a key solution to support the energy transition. This work presents a predictive control framework specifically designed to enhance the sustainability, autonomy, and intelligence of hybrid microgrids integrating lithium-ion batteries and hydrogen-based storage systems.
The proposed control strategy focuses on day-ahead optimisation of energy dispatch over 24-hour horizons. It aims to maximise renewable self-consumption, reduce greenhouse gas emissions, and ensure system autonomy under varying operating conditions. The hybrid storage configuration includes fast-response batteries and a long-duration storage layer based on an electrolyzer and fuel cell, allowing the system to efficiently store surplus solar energy and reuse it in low-generation periods, fully avoiding fossil-based backup systems.
Unlike conventional economic dispatch methods, the control algorithm is built upon a stochastic Model Predictive Control (SMPC) architecture with a multi-scenario formulation. It considers uncertainty in solar generation and potential grid disconnection events, allowing the system to anticipate and adapt to both normal (grid-connected) and critical (islanded) scenarios. A Mixed Logical Dynamic (MLD) model enables the incorporation of logical switching, component-specific constraints, and auxiliary variables to represent system behaviour more realistically.
What differentiates this approach is its emphasis on autonomous and intelligent decision-making. The controller dynamically selects the most appropriate energy sources based on forecasts and internal state, while also minimising operational stress and component degradation. For example, it avoids inefficient start-stop cycles in the hydrogen subsystem by using penalisation terms within the objective function. This proactive behaviour reduces energy waste, prolongs asset life, and enhances the long-term sustainability of the system.
Additionally, the control logic embeds resilience criteria to guarantee the supply to critical loads, such as those in hospitals or public infrastructure, even during extended islanded operation. These constraints ensure that energy resources are allocated not only based on cost, but also on availability, reliability, and environmental impact.
Simulation results using a hospital-type microgrid in MATLAB/Simulink confirm the benefits of this integrated control approach. The system increases renewable self-consumption, reduces curtailment, and maintains a reliable supply during grid outages. The MPC-based controller reacts to forecast deviations and re-optimises dispatch in real time, demonstrating its suitability for real-world scenarios with fluctuating demands and generation. In addition to technical performance, the control system contributes to emissions reduction and energy independence, aligning with climate targets and sustainability goals.
The modularity and scalability of the proposed control architecture make it suitable for deployment in a variety of settings, including critical infrastructure, research campuses, and isolated communities. By combining intelligent automation with renewable integration and storage coordination, this approach enables microgrids to function as fully autonomous, low-emission energy hubs capable of supporting the wider transition to green and resilient energy systems.
