Enemy position prediction study for wargame based on CNN-SALSTM

With the development of military research, enemy position prediction in wargame has become a key task. In this paper, a CNN-SALSTM model is constructed for enemy position prediction based on the open-source land wargame prediction dataset from the Institute of Automation, Chinese Academy of Sciences. The model combines CNN, LSTM and self-attention mechanism to effectively capture data features and relationships. By comparing with CNN-GRU and CNN-LSTM, the results show that the CNN-SALSTM model has higher accuracy and lower loss value, and has better top-x accuracy in the training set and test set, especially in the top-10, which verifies the validity of the model, and provides a new method for the prediction of enemy position in wargame.