Target Operator Trajectory Prediction Method Based on Attention Mechanism and LSTM

Abstract

Aiming at the problem of incomplete information in wargames, this paper uses the encoding-decoding model of Long Short-Term Memory (LSTM) to divide the target trajectory prediction into two decoupling processes, encoding and decoding, using the decoding process Output the parameters of the learning probability distribution for trajectory prediction, and introduce an attention mechanism. Two attention layers are added to the LSTM model and the realization principle of the attention layer is introduced in detail. The context information is enriched to achieve the improvement of the model, and finally the simulation analysis The effectiveness of this method is verified.