Post-War Assessment Method for Over-The-Horizon Air Combat

With the development of air combat training towards actual combat, the need for post-battle training evaluation of over-the-horizon air combat is becoming increasingly urgent. A post-battle evaluation method for over-the-horizon air combat is proposed using deep learning techniques. A long short-term memory network (LSTM) is proposed to train a win-rate prediction model for real-time assessment of air combat situations. The Otsu method is used to determine the threshold value of the key time node, and the moment when the absolute value of the difference in win rate is greater than the threshold value is taken as the key time node. Training evaluators can only evaluate the operational training process by playing back the training data output by the method at the key time node, which makes the post-battle evaluation much more efficient. The experiments show that the deep learning network predicts win rates accurately and that the key nodes are determined accurately.