Aiming at the different threat levels under phased array radar multi-target tracking， the Bayesian Cramer-Rao lower bound （BCRLB） of the target position estimation is used as the allocation criterion. The paper establishes a multi-target tracking time resource allocation optimization model based on the threat degree. The model based on the threat degree to track the target can be divided into two categories and different types use different time resource allocation methods. Due to the time-consuming operation and optimization algorithm， this paper also proposes a multi-target tracking time resource fitting method based on back propagation（BP） neural network. Computer simulation shows that the model and the method can keep the target tracking in the best state， and the BP neural network reduces time consumption by more than two thousand times.