The Research On Decision Information Fusion Based On Rough Set And Evidence Theory

In the information fusion, the information is always in the face of incomplete, imprecise or uncertain. Rough set theory and evidence theory are important methods of information fusion to deal with the uncertainty. Rough set approximation space is used to deal with uncertain or imprecise knowledge. Moreover, this method only needs the existing knowledge, and does not require any priori information. So it can be more objective when dealing with uncertainty. However, rough set is weakly on reasoning. Evidence theory is an important tool for uncertainly reasoning. The trust interval can be built by the belief function and the plausible function, to measure the degree of uncertainty of the evidences. Therefore, evidence theory can be used to represent the uncertain information.Based on the complementary relationship of rough set and evidence theory, the information fusion by the use of rough set and evidence theory is researched. The work is mainly about how to access evidence in the different trust structure and synthesis. Because of the different focal element, the trust structure is different. Single-point set, the partition and the cover of the domain, these three trust structures are special. Because of the different trust structure formed by belief functions, the process of obtaining evidence is different. These three special trust structures are discussed. And the relationship between evidence theory and probability theory, between evidence theory and rough sets is researched. Then how to obtain evidence function under the different trust structure by rough set is considered.Another key issue of information infusion is the synthesis of the evidence. On the basis of the architecture of rough set and evidence theory, the difference of the synthesis of the evidence between the rough set and evidence theory is analyzed in this thesis. And the relationship between the division of the partial sequence stratigraphic and the combined operations of the evidence theory is studied. It analysis different divisions in the partial sequence stratigraphic by the use of granularity. The supremum, infimum, the upper bounds and lower bounds of the partial sequence stratigraphic export different partitions of the domain. The mass functions obtained from these partitions don’t correspond to the synthesis of the mass functions in evidence theory. Therefore, it clarifies the relationship between the mass function obtained from the partitions in rough set and the synthesis of the mass functions in evidence theory.