Target Intention Inference Model Based on Variable Structure Bayesian Network

Target intention inference is an important aspect of situation assessment. The evidence system of targets´ intention inference is discussed according to the independent relationship between targets´ intention and input evidence. The targets´ intention probability inference model is proposed based on static Bayesian network. In order to expand the application domain and predigest the parameter learning contents, the decomposition and mergence of network´ structure are disposed. The process of parameter learning is simplified according to the condition independent relationships. Different network architecture and condition probability state space of their parameter learning method are carried on. The result shows that variable structure is a suitable method for reducing the state space of the network conditional probability.