Research And Application Of Bayesian Game Information Fusion Model And Algorithm

This thesis supported by the National Natural Science Foundation.Under combing Game Theory and Information Fusion,it presents an information fusion model which is called Bayesian information fusion model. Multi-source information fusion that lies in the conflict environment is solved based on the model. Bayesian network structure is constructed based on the model,and information game fusion is achieved by the Multi-tree propagation algorithm which belongs to Bayesian network algorithm.Finally,the fusion and algorithm is applied to ETC,and to solve the problem about the accurate of vehicle identification,Then the validity of ETC is improved.●Starting from the basic theory of this paper, the functional model of information fusion and some fusion algorithms are introduced based on the introduction of definition, and the level of fusion.As the information game exist in the environment of competition and cooperation, three elements of game model are introduced in game theory, and strategy game and extended game are introduced detailly. From these basic theories, summarized the basic idea of game model, and the game information fusion model is established.●Bayesian game is mainly used to solve the game problem in uncertain environment.Bayesian network is built firstly. Then game fusion expressed by Bayesian network structure.Specific information fusion is achieved through multi-tree propagation algorithm which belongs to Bayesian network algorithm. The prior probability is amended by Bayesian probability model in the fusion process.●In practice, the model and algorithm is applied in the freeway electronic toll collection system, to solve the problem about the accurate of vehicle identification.●Game information fusion theory is not only applied in ETC, and more widely used in pattern recognition and artificial intelligence. Keywords:Information Fusion; Bayesian Game; Vehicle Identification...