Due to complex structure, flexibility and strong mobility of UAVs swarm, the existing time registration model of Cooperative Tracking can not effectively perform the time error compensation task for the trajectory of drone swarm. In this paper, clustering and approximate entropy are used as preprocessing steps in the auxiliary decision making of anti-UAVs swarm system firstly, then BP network and Kalman filtering are combined to complete the time compensation task. Experiment results show that, based on the improved auxiliary decision making method, the adaptive time registration model can perform better in accuracy and speed compared with the method fully using Kalman filtering.