Research On Air Target Intention Recongniton Based On Multi-entities Bayesian Network

In the condition of information-based war, air target intention recognition is faced with some new problems. On one side the air targets become formidable (containing intention recognition) in area defense system because of high technology, on the other side kinds of advanced reconnaissance equipments and computer artificial intelligence methods provide new methods for the air target intention analysis. Traditional air target intention recognition methods pay attention to the data processing and application, but pay less attention to the function of air target action rule in intention recognition. So these methods guadually cann't meet requirements of air target intention recognition in information age. So the research on air target intention recognition has important practical significance and application value.As to this problem, this paper present air target intention recognition based on Multi-Entities Bayesian Network (MEBN). The main contributions are listed as follows:(1) An air target intention recognition anylysis frame is proposed based on MEBN. Firstly Multi-Entities Bayesian Network basic theory and First-Order Logic (FOL) theory are dissertated firstly. Secondly a simple scenario and air target intention recognition uncertain factos are anylyzed. The configuration and function of the frame are dissertated by intention space, air target action rules and air target entity models. And then the air target intention recognition course based on MEBN is proposed. Especially the important effect of action rules in air target intention recognition is brought into prominence. Lastly the key techniques are anylyzed.(2) The air target action rule model ActionModel(f₁ ,f₂, , f_n)e(E,B,A,C) based on FOL is presented.The concept of air target action rule is recapitulated. From the view of the air target executes task, and following the air target control and command course, the air target action rules classification is proposed. The consistency of the model is disscused, and an action model example is dissertated. ActionRul(3) The air target entity model and intention recognition method based MEBN are proposed.The characters of the air target entity object are described as MEBN fragments, the air target entity model is proposed based on the MEBN Theory. And the air target intention recognition based on MEBN is dissertated in detail from the three aspects of intention recognition course; Bayesian Network generated method and MEBN inference mothed.(4) At last, the above-mentioned research is analyzed and validated by an air defense scenario. The result verifies the validity and the feasibility of the method.