The prediction of air target trajectories is important. It is necessary to integrate time-series data in the motion of air targets, such as timestamps, spatial coordinates, kinematic parameters, etc. To cope with the challenges of air target trajectory prediction in complex electromagnetic environments, this paper proposes the MSDTransformer model. The SMSAE and DDE modules are introduced, and this method improves the performance of air-target trajectory prediction. The sensitivity analysis of system hyperparameters verifies the rationality of model parameter configuration. The experimental results show that, among the simulated air target data, MSD-Transformer achieves lower prediction errors and demonstrates strong overall performance, compared with 5 main methods (iTransformer, Informer, PatchTST). It can support the understanding and prediction of target situations, provide an accurate basis for the analysis of the intentions of air targets and the prediction of the evolution of situation evolution, and then improve the situation awareness ability. It has important application value for trajectory prediction in modern aerospace surveillance and situation awareness.