Abstract: During the operation of twostep cold crucible glass solidification device, the pressure control of cold crucible and calciner is very important. However, due to the periodic decomposition or sealing problem of calcined products, there will be large pressure fluctuation. If the pressure control is insufficient, it is easy to cause leakage or bring excessive radioactive substances into the tail gas system, resulting in excessive emissions. Therefore, aiming at the problems of large fluctuation, nonlinearity, large delay and uncertainty in the negative pressure control of glass solidification cold bench of HLLW based on cold crucible, an adaptive negative pressure control system was designed by using fuzzy algorithm. The fuzzy algorithm established in this system can predict the possible pressure fluctuation according to the collected data. Combined with PID control, the negative pressure can be controlled by adjusting the electromagnetic air supplement valve. The simulation model of the adaptive negative pressure controller was established through the Simulink module of MATLAB, and the historical operation data were used to verify whether the air replenishment calculated by the adaptive negative pressure regulation system can maintain the negative pressure stability in the calciner. The results show that the accumulation of negative pressure deviation is reduced by 98.64%, and it is proved that the adaptive negative pressure controller can effectively reduce the pressure fluctuation and eliminate the hidden danger of leakage.