The pressure stability of the boil-off gas (BOG) system in liquefied natural gas (LNG) terminal is crucial for ensuring the safe and economical operation. However, the proportional integral derivative (PID) control strategy commonly used in LNG terminal cannot adapt to the characteristics of large time delays, strong variable coupling, strong disturbances, and large-scale operating conditions, which has led to significant pressure fluctuations and frequent manual interventions in the BOG system. Therefore, a multi model prediction method with feedforward compensation was proposed. Firstly, based on analysis of the process and variable correlation, the BOG generated by unloading was determined to be introduced as a feedforward variable into the controller, forming a feedforward feedback composite control structure to cope with strong system disturbances. Furtherly, the impact of tank liquid level was identified, thus multiple dynamic matrix models were established for different liquid level intervals, and a hysteresis model switching strategy was developed to solve the problem of model mismatch in large-scale operating conditions. In addition, an efficient shaping optimization algorithm was adopted to overcome the calculation problem caused by the discrete numerical value of the compressor load, and to achieve the calculation of the compressor load distribution and start stop control strategy under the storage tank control objective. Taking a LNG receiving station in China as an example, through experimental simulation and comparison with non feedforward and single model control schemes, it is determined that the proposed method has significant advantages in key control indicators such as overshoot, stabilization time, and pressure variance. The proposed feedforward multi-model predictive control model has excellent control performance in the LNG storage tank BOG compressor system, solving problems such as strong coupling, large-scale operating condition switching, and multiple interference factors. It effectively improves the pressure stability of the BOG system and reduces the operating energy consumption of the BOG compressor, providing strong technical support for practical engineering applications.