This paper delves deeply into the problem of automatic calibration in the thermal hazard simulation of full-scale automotive climate numerical wind tunnels. The vehicle-level simulation model for automotive thermal management contains a large number of parameters, and it is difficult to obtain accurate values, making the parameter calibration work crucial. To this end, this paper constructs an automated optimization process and adopts a dual-track technical approach: one is manual calibration based on the comparison of experimental data and expert experience, and the other is intelligent calibration with the help of multiple technical means. At the key technology level, multiple software nodes are integrated through graphical software to build a parametric simulation process; various Design of Experiments (DOE) algorithms such as Sobol sampling and full-level sampling are integrated; sensitivity analysis is carried out to clarify the sensitivity ranking of input parameters; artificial intelligence algorithms are used to generate and train surrogate models, which significantly reduces the simulation calculation time. The above methods ensure that, on the premise of accurately loading the simulation boundary conditions, the errors of temperature and wind speed at the corresponding positions of the simulation model and the test measurement points are controlled within the engineering acceptable range, significantly improving the accuracy and reliability of the simulation results of automotive thermal management.