This project introduces Dynamic Bayesian Networks (DBN) and Sequential Bayesian Networks (SBN) to construct a DSBN model that describes tactical behavior recognition and inference for the dynamic sequentiality problem of tactical behaviors of enemy targets in complex battlefield environments. We also introduce probability trans-fer-MFRags (PT-MFrags) and seriesrelation-MFrags (SR-MFrags) in the algorithm, and extend MEBN to have the ability to express the migration process and sequence process of event states, which supports the construction of DSBN model, support DSBN model construction. The simulation results show that by combining the advantages of DBN and SBN, the model is able to deal with time series data and dynamically changing situational information, which improves the accuracy and timeliness of prediction, which can be more flexibly applied to different battlefield simulations, thus promoting the improvement of the corresponding tactical level, and maximizing the combat effectiveness of UAV formations.