With the rapid development of hydrogen energy and medical diagnostics

there is a growing demand for high-precision and highly robust hydrogen sensors.To address the limitations of nickel oxide（NiO）-based hydrogen sensors

including poor selectivity

high operating temperature

and susceptibility to environmental interference

a co-optimization strategy of materials and algorithms is proposed.Graphite-carbon-encapsulated NiO composite nanosheets (NiO/C-NS) are successfully synthesized via a gel blowing and pyrolysis meth

od.The resulting sensor exhibits high sensitivity(response value of 6.24 at 8000×10

-6

hydrogen)

excellent selectivity(response ratio of hydrogen to carbon monoxide reaching 12.8)

and good stability at a low operating temperature of 73℃ in the hydrogen concentration range of 100×10

-6

~8000×10

-6

.Furthermore

a deep adaptive graph neural network(DAGNN)is developed to effectively suppress signal drift caused by temperature and humidity variations.Under 32 different environmental conditions

a mean relative error(MRE)of 4.9% and a coefficient of determination (

R

2

) of 0.99 are obtained in hydrogen concentration prediction

significantly outperforming conventional methods.A complete design paradigm covering sensitive materials to intelligent algorithms is provided in this paper for developing high-performance and high-robustness hydrogen sensors suitable for industrial safety and new energy applications