Trajectory prediction for an incoming missile via intent recognition and guidance parameter identification

The accurate trajectory prediction of an incoming missile enables defense systems to effectively neutralize potential threats, thereby protecting civilian populations, military personnel, and infrastructure. Current researches focus on prediction under the assumption of knowing the attack target at the beginning of the engagement, which is seldom the case in reality. To deal with this issue, an intent recognition model based on a Gated Recurrent Unit (GRU) neural net-work is proposed in this paper. The inputs of the network are the available measurement information between the target and incoming missile, while the outputs are one-hot labels. To increase the training speed of the network and enhance its generalization capability, the adaptive moment estimation (Adam) algorithm is adopted for the training process. Based on the information from the network, a cubature Kalman filter (CKF), which integrates a higher-degree cubature rule to approximate the state distribution of the nonlinear dynamic system, is introduced to estimate the state and predict the trajectory of the incoming missile. Simulations present the transition process of the network and demonstrate that the proposed method achieves faster convergence and higher prediction accuracy compared to traditional approaches that are solely based on Kalman filters.