Target Intent Recognition Method Based on Evidence Fusion in TimeSeries Networks

Target tactical intent identification in beyond-visual-range air combat is investigated by a novel method of target intent recognition based on information fusion, which combines the advantages of Dempster-Shafer evidence theory and deep temporal networks. The first is by constructing a 1DCNN-BiLSTM deep temporal network to extract the target change features in terms of trajectory and situation; the weighting coefficients of the evidence are proposed to be generated using information entropy of both kinds of evidence as the basis of weighted discounting so that the credibility of the proof is improved, and finally a more reasonable intent fusion result is obtained. The method proposed in this paper is applied to the test of the actual antagonistic data, and the result shows that the method has good dynamic performance and classification accuracy.