Target Tactical Intention Recognition Based on Fuzzy Dynamic Bayesian Network

In order to make full use of the advantage of fuzzy set theory in discretizing continuous variables and reduce the uncertainty brought by traditional static Bayesian network, this paper applies the method of fuzzy dynamic Bayesian network to reasoning learning of battlefield target tactical intention recognition based on the analysis of the observation values of height, speed and distance. The simulation results show that the method is effective and can provide new ideas for target tactical intention recognition. Introduction In view of situation assessment, the Joint Council of Laboratories of the United States Department of Defense (JDL) proposed a multi-level hierarchical battlefield information model and a relatively recognized definition of battlefield situation assessment [1], which has an important impact on understanding situation assessment, and also provides a reference for scholars in various countries to carry out relevant research. Target tactical intention recognition[2] has always been a research difficulty in this field, mainly because there are many uncertain factors in target intent recognition. Bayesian network[3] is one of the most effective probabilistic relational image description models in uncertain knowledge representation and probabilistic reasoning. Professor Pearl establishes the basic theory system of Bayes network[4], uses the characteristics of Bayesian network to gather and identify, and determines the direction of some edges based on Bayes statistics and graph theory. Traditional Bayesian networks[5] refer to static Bayesian networks, which do not provide a direct way to express time dependence. Dynamic Bayesian network[6] (DBN)adds time dimension to traditional static Bayesian network. In addition, because the reasoning and learning process of continuous Bayesian network is more complex, and in practical application, Bayesian network of continuous nodes and Bayesian network of mixed nodes are widespread. In order to make full use of the advantage of fuzzy set theory in discretization of continuous variables, the clear node variables of dynamic Bayesian network are extended to the fuzzy node variables. The method of fuzzy dynamic Bayesian network is applied to reasoning learning of battlefield target tactical intention recognition. Mathematical Description of Tactical Intention Recognition Tactical intentions of targets are hidden in specific actions or behaviors of targets, and can not be observed directly. Therefore, the process of inference of target intentions can be carried out according to the information acquired, the principles of tactical use, the methods of use and the commonly used domain experience knowledge, combined with the observed target actions and behavior patterns. Assuming that the knowledge of military field is MK MK ,MK ,...MK and the real-time data information is RD RD , RD ,...RD , the estimation of tactical intention can be described as the determination of confidence P H|K, S of uncertain tactical intention TI TI , TI , ... , TI , where TI is the target tactical intention space constructed, and TI , TI , ... , TI is a division of space TI, i.e. International Conference on Modeling, Analysis, Simulation Technologies and Applications (MASTA 2019) Copyright © 2019, the Authors. Published by Atlantis Press. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). Advances in Intelligent Systems Research, volume 168