Projectile trajectory and launch point prediction based on CORR-CNN-BiLSTM-Attention model

Aiming at the problem that it is difficult to balance accuracy and the time length of projectile flight trajectory prediction, this paper combines the advantages of bidirectional long short-term memory network (BiLSTM), convolutional neural network (CNN) feature extraction and attention mechanism (Attention), and proposes a trajectory prediction model with correction (CORR-CNN-BiLSTM-Attention). For the end error of 20 s, there are only 7.8 m, 7.9 m and 0.9 m deviations in firing height, range and offset. The network structure can ensure the prediction accuracy and increase the prediction time length. This scheme can provide sufficient response time for missile interception and effectively improve the probability of missile interception. At the same time, a network structure proposed in this paper trains two models and uses them respectively for future trajectory prediction and reverse launch point prediction. This scheme can achieve accurate prediction. Among them, the comprehensive error of the launch point of the reverse prediction sea level height in the range and sideslip direction is 8.31 m, which can accurately predict the position of the enemy launch point and strike the launch point.