Research On Situation Assessment Techniques In Information Fusion

With the rapid development of science and technology, especially the informationtechnology like sensor, communications, computer, etc, Information Fusion technique is beingwidely used in military, finance, biology, and other fields. As an important question for studyin modern scientific research, the study of Information Fusion technique aims to optimize andfuse information from multiple sources, in hopes that the incidents can be described andassessed more objectively, and future tendency of the situation can be predicted accurately. Ininformation fusion technique, Situation Assessment is a critical procedure and of greatsignificance. The thesis does a systematic research on Situation Assessment theories andmodel, reducting of Assessment rules, and Situation forecasting. The contents of the thesis areoutlined as follows:1. The thesis first analyzes the realization of Situation Assessment and establishes themodel of multiple levels according to the functions fulfilled by Situation Assessment and itsthree processes—situation perception, situation interpretation and situation prediction. Bystudying the magnanimity of the knowledge about situation, a method about knowledgeacquiring is proposed. Based on the analysis of Situation Assessment outline, the templatemodel based on event-driving are proposed in the light of its different processes andinformation sources.2. Analyzing the mass and uncertainty of information in the situation perception, andmaking use of the knowledge discovery methods based on Rough Sets, the thesis thenproposes the model of binary-discernibility-matrix so that a queue based on the number ofelements is constructed and searching space in the process of knowledge reduction is reducedin size. The efficiency of knowledge reduction is improved.3. For the rule-reduction problems of knowledge database and by further studying thereduction of Rough Sets knowledge, an Attribute-Value-Tree model is set up; By making bestuse of tree structure and tree mergence algorithm, a new knowledge reduction algorithm iscreated to reduce the time and space complexity of Attribute Reduction and Value Reduction,and to increase the efficiency of matching rules.4. After studying the theories of Plan Recognition in Situation Assessment, SituationAssessment is proved to be a process of Plan Recognition. The disadvantages of currentmethods of Plan Recognition lie in its ambiguity and uncertainty of knowledge. This thesisdiscusses a new method of Plan Recognition on the basis of Bayesian Networks and putsforward an inference model of Bayesian Networks on the basis of Intuitionistic Fuzzy expression so as to overcome the contradiction of old method. An example of the intelligentplanning of military tactics is further analyzed to introduce a situation assessment model ofBayesian Networks inference based on Intuitionistic Fuzzy expression.5. The thesis finally discusses the influence of uncertain factors in Situation Assessmentand proposes a model of Situation Assessment information based on Intuitionistic Fuzzyexpression. A method of situation assessment based on multi-attribute Intuitionistic FuzzyDecision is also given. An example further illustrates the process of situation prediction,indicating the feasibility and efficiency of the method in solving situation assessmentproblems.