Research And Application Of Evidence Theory And Convex Function Evidence Theory Based On Gaussian Function

Evidence theory as a most commonly used uncertain and inconsistent method ininformation fusion field has been widely used in many fields. But the evidence theory indealing with high-conflict evidence has a paradox, and the integration of the results of thistheory in dealing with the orderly proposition class of problems often violates common senseand appears bimodal fusion results. These problems led to the application limitations of thetheory of evidence combination rules. So, the conflict between the evidence theory Paradoxand evidence, facing the orderly proposition class of problems of convex evidence theory basedon Gaussian method and its application of making the evaluation of multi-attribute decisionproblems is made research as following：This paper studies the relationship between the conflict factor K value and the degree ofconflict in the evidence and the impact of the synthesis results the mass difference between thevalues, derives that K value are not well reflected of the degree of conflict between theevidence, also pointed out that the evidence of different focal the differences of belief function(mass) are closely associated with trust convergence. The article introduces a parameter ofdifferentiation factor, presents a new method of evidence combination, experimental resultsshow that the method can deal with conflict evidence integration.This paper introduces the concept of ordered propositions, and describes in detail themodel of convex evidence theory which deal effectively with the orderly proposition classproblems. So far, the model is still the only processing orderly proposition uncertaintyreasoning. But for some basic support function value, convex function evidence theorycombined result is not satisfactory, this paper proposes a Gaussian function instead ofintegrated function of convex function evidence theory model, it is a better way to solve theabove problems.This paper points out multi-attribute evaluation problem is an ordinal type of propositionproblem areas, so you can use the Gaussian convex function of evidence theory model to solvethe problem. In this paper, on the basis of the Gaussian function convex evidence model, usingthe method based on typical samples and the method based on AHP method to build basicsupport functions, and thus is given two Gaussian convex function evidence theory modelbased on typical samples and based on AHP. Among them, the Gaussian model convexfunction evidence theory based on typical sample using normal curve to generate typicalsimilarity, then obtain the basic support functions of the test samples to the targetpattern,Finally use the Gaussian function convex evidence model Synthesis of basic support functions to obtain comprehensive results; the Gaussian function convex evidence model,using the method based on AHP first determine the levels of indicator’s assurance,then useAHP to determine index weight, and calculated for the basic belief function of each indicatorevaluation objectives, use the Gaussian function convex evidence model Synthesis of basicsupport functions to determine the target evaluation level.