The Research On Decision Support System Based On Evidence Theory

D-S evidence theory is an important methodology used for uncertain information. Because of the usage of interval estimating in modeling uncertain information, evidence theory has many advantages, such as the flexibility for information modeling, and it's used more and more in many fields such as uncertainty reasoning, group decision making, artificial intelligence, multi-criteria decision making and etc, since it was introduced first.In this dissertation, an uncertainty reasoning approach based on evidence theory is introduced to decision support system for disaster relief management. Some key technologies in this approach, such as evidence relevance mechanism, conflict-resolving methods, and representation for uncertain knowledge, establishment of knowledge base based on uncertain rules and uncertain reasoning method among rules which are based on evidence theory and etc, were studied in detail.Main contents are as follows:(1) Based on the introduction to the evidence theory and its newest advances, the basic evidence theory's concept, methods, existing problems and its solutions are discussed in detail.(2) Aiming for evidence relative problem, based on evidence relative intensity handling mechanism, a relative independent evidence method is used and, experiment results show its validity.(3) For solving evidence conflict problem, similarity among evidences is introduced and a conflict-resolving method based on evidences'similarity is proposed. Experiment results show its validity(4) On the basis of discussion of framework of evidence reasoning system in decision support system, using uncertain knowledge's representation and uncertain rules to build uncertain knowledge base and evidence reasoning machine. An evidence rule reasoning approach based on Rete net is given and its reasoning process is listed in detail.