Deep learning-enhanced techno-economic optimization of hybrid wind-solar-hydrogen system for Dutch heating networks

This study presents a comprehensive techno-economic-environmental analysis of integrating hydrogen into Dutch heating networks to enhance seasonal storage and reduce CO₂ emissions. A hybrid solar–wind–hydrogen system is proposed, comprising photovoltaic panels, wind turbines, a battery energy storage system, a proton exchange membrane electrolyzer, hydrogen compression and storage units, a proton exchange membrane fuel cell, and a water-to-water heat pump. An adaptive peak-shaving controller is developed to govern the battery energy storage system, dynamically limiting grid import peaks while directing surplus renewable electricity to hydrogen production; its application results in a 72 % reduction in the highest observed peak. The system is tested for the city of Enschede, where the controller enables the battery to support peak shaving and coordinates the routing of excess renewable electricity toward green hydrogen generation, which is stored and later used in cold spells. A deep learning-assisted optimization framework, combined with a genetic algorithm, significantly reduces computational costs while accurately predicting system performance. The results show that hydrogen enables seasonal storage, achieving an exergy efficiency of 35.04 %, a total cost rate of 4.84 €/h (5.24 $/h), annual CO₂ emissions of 63.64 tons, a levelized cost of hydrogen of 6.48 €/kg, and a 42.21 % share of the heat supply mix during cold spells.