Prediction of Battlefield Target Trajectory Based on LSTM

Abstract

Battlefield situation prediction is an indispensable part of battlefield situational awareness, and refers to one of the important references for tactical decision-making by warfighters. The battlefield situation prediction is to use the information obtained in the two phases of situation awareness and situation recognition in battlefield situation awareness to predict the battlefield situation, and obtain the changing trend of the enemy's target state, position, and intention. Aiming at the temporal correlation of Long Short Team Memory Network (LSTM), this paper proposes to build a target trajectory prediction model based on LSTM network. First, the motion trajectory information obtained in the two stages of situation awareness and situation recognition is normalized and time-series processed to construct a time-series sample data set; then, the time-series data set is pushed to a target trajectory prediction model based on the LSTM network. Training; Finally, use the trained model to complete target trajectory prediction. The experimental results show that the prediction method using the LSTM model has the characteristics of high accuracy and easy implementation, and it has obvious advantages in processing sequence data.