Tongxin xuebao (Mar 2023)
Conflict evidence combination method based on clustering weighting
Abstract
Aiming at the failure of the combination of highly conflicting evidence, the combination of the relationship between evidences and the characteristics of evidence itself was considered, and a method of conflicting evidence combination based on clustering weighting was proposed.Firstly, the similarity coefficient was constructed to cluster the evidence, and the inter-class weight was determined by class capacity to reflect the differences among classes.Then the information entropy was used to obtain the intra-class weight of each evidence.The modified weight of the evidence was obtained by combining the inter-class weight and intra-class weight.Finally, the conflict evidence was modified according to the weight, and the combination rule of evidence theory was used to achieve the final synthesis of the modified evidence.The results of the example show that the proposed method can reasonably carry out evidence correction, and the combined results have better focusing effect than other methods.
