A BILSTM+Attention Target Intention Prediction Method Considering Threat Assessment
Qiuni Li, Zhenyu Huang, Chaozhe Wang, Zongcheng Liu · 2025
This paper proposes a multi-feature continuous time-series BiLSTM+Attention three layer air combat intent prediction method that includes trajectory prediction, threat assessment and intent prediction for predicting the tactical intent of target aircraft in air combat. To prevent the intent prediction results from being biased by single moment state information, this method considers predicting from multiple state features and trajectory information in continuous time-series. First, an LSTM neural network with excellent memory function is used to predict the trajectory of the target aircraft. Then, threat assessment of the target aircraft before and after the prediction is carried out based on the trajectory prediction result. Finally, a BiLSTM+Attention model with bidirectional correlation is used for target intent prediction. By comparing the four different algorithms of BiLSTM+Attention with threat assessment and BiLSTM+Attention without threat assessment, BiLSTM, and LSTM in this paper, the experimental results show that the proposed method has the best performance in terms of accuracy, precision, recall rate and F1 score value, with the best prediction effect.