A systematic review of predictive, optimization, and smart control strategies for hydrogen-based building heating systems

The use of energy in the built environment contributes to over one-third of the world's carbon emissions. To reduce that effect, two primary solutions can be adopted, i.e. (i) renovation of old buildings and (ii) increasing the renewable energy penetration. This review paper focuses on the latter. Renewable energy sources typically have an intermittent nature. In other words, it is not guaranteed that these sources can be harnessed on demand. Thus, complement solutions should be considered to use renewable energy sources efficiently. Hydrogen is recognized as a potential solution. It can be used to store excess energy or be directly exploited to generate thermal energy. Throughout this review, various research papers focusing on hydrogen-based heating systems were reviewed, analyzed, and classified from different perspectives. Subsequently, articles related to machine learning models, optimization algorithms, and smart control systems, along with their applications in building energy management were reviewed to outline their potential contributions to reducing energy use, lowering carbon emissions, and improving thermal comfort for occupants. Furthermore, research gaps in the use of these smart strategies in residential hydrogen heating systems were thoroughly identified and discussed. The presented findings indicate that the semi-decentralized hydrogen-based heating systems hold significant potential. First, these systems can control the thermal demand of neighboring homes through local substations; second, they can reduce reliance on power and gas grids. Furthermore, the model predictive control and reinforcement learning approaches outperform other control systems ensuring energy comfort and cost-effective energy bills for residential buildings. • Explore hydrogen role in decarbonization of residential heating systems. • The study outlines merits and demerits of switching to hydrogen for heating systems. • Semi-decentralized heating systems have the potential for improved manageability. • Machine learning models help manage the randomness in residential building datasets. • Optimal control plans contribute to maintaining thermal comfort and energy savings.