The hierarchical SVDDBNS based on modularization concept for air target recognition

The process of Air target identification is hierarchical, and is also a process of data fusion of diversified information obtained in the unstable time domain. In this paper, the process of air target recognition is regarded as a process of a qualitative inference. According to the features that the hierarchy of air target identification process and the input parameters obtained in the unstable time domain, we constructed the air target recognition model on hierarchical structure-varied discrete dynamic bayesian networks (hierarchical SVDDBNs) by modularization concept. The air target identification model has such features, that is , the constructed model can real-time reconstructed the networks and finish the tasks flexibly by the features of the input data. In constructing bayesian networks model, the changes of structure is regular, in addition, the number of network nodes don´t influence the decouple of the state of network nodes each other. Such can avoid structure learning and parameters learning. In the paper, the inference algorithm is presented, and simulation results show the feasibility of this approach.