HyTwinOpt

HyTwinOpt aims to develop a framework for multi-objective modelling, control, and optimization of integrated hydrogen energy systems (IHES) operating under stochastic renewable and dynamic grid conditions. The project addresses the challenge of coordinated operation of hydrogen production, storage, and conversion within complex multi-energy systems integrating renewable energy sources (RES), grid interactions, and battery storage.

The central innovation is a physics-informed, machine learning enhanced Digital Twin, providing a high-fidelity predictive and decision-support environment that bridges first-principles models and data-driven methods. This approach captures nonlinear system dynamics and mitigates uncertainties arising from RES variability and demand fluctuations.

A Hybrid Energy Management System (H-EMS) will combine forecast-aware scheduling, machine learning-based prediction, and finite-state machine (FSM) control to ensure robust and adaptive operation. The methodology spans component-level modelling, ML model training, and full Digital Twin integration, supported by advanced optimization methods such as Model Predictive Control (MPC) and Reinforcement Learning (RL).

The framework will be validated on a laboratory-scale hydrogen microgrid under grid-connected and islanded conditions. Expected outcomes include improved system efficiency, reduced energy losses, enhanced reliability, and extended component lifetime.

The project advances the field by establishing a unified, system-level methodology, shifting research from isolated component optimization toward integrated, predictive, and adaptive control of hydrogen energy systems. HyTwinOpt thus provides a scalable pathway for large-scale deployment of hydrogen technologies, supporting the transition to low-carbon power systems and sustainable energy independence.