With the widespread application of hydrogen energy as a clean energy source, its safety issues are increasingly being emphasized. Hydrogen leakage can lead to fires or explosions, posing a threat to public safety. Conducting safety research on the diffusion patterns of liquid hydrogen leakage in response to this safety issue can help prevent and reduce hydrogen leakage accidents and protect people's lives and property. This study employs the computational fluid dynamics (CFD) software Fluent, in combination with the Lee model and the VOF model, to simulate the leakage and vaporization process of liquid hydrogen. It is found that an increase in wind speed extends the downwind diffusion distance and reduces the vertical height. Based on the diffusion patterns of the combustible gas cloud and in view of the limitations of hydrogen sensors, a new method of using temperature sensors to monitor environmental temperature changes to predict hydrogen concentration is proposed. By analyzing the mathematical relationship between temperature and hydrogen concentration during the initial stage of leakage, a mathematical model is established. The validation of the model shows that temperature sensors arranged on the 0.8 m height plane can quickly respond to leakage events, thereby enhancing the safety management level of hydrogen refueling stations.