The cryostat equipped with superconducting materials is one of the most important component of high-temperature superconducting maglev vehicles. However, due to the multiple energy disturbance forms in the cryostat, such as AC losses of superconductors, vibration of the cryostat and so on, the evaporation of liquid nitrogen (LN2) will be stimulated to accelerate, which will seriously threaten the vehicle safety. Therefore, this paper designs an intelligent controller which can realize the function of auto-filling LN2 based on the Kalman filter (KF) algorithm. Moreover, in this publication the corresponding methodology will be described as well in detail. Initially, the platinum resistance sensors were used to study the relationship between the temperature and LN2 level, meanwhile, the system state equation and observation equation were established. Then, the KF algorithm was introduced to estimate the true LN2 level. Based on the feedback signal of LN2 level, the system utilizes the STM32 microprocessor as control chip to control the solenoid valves and realizes the function of auto-filling. The controller has been implemented and the total time for filling one cryostat was reduced from 86 min to 16 min. This result is very encouraging and inspiring for the practical application of HTS maglev systems.