A Novel Fuzzy Approach for Combining Uncertain Conflict Evidences in the Dempster-Shafer Theory

Abstract

Fusing conflict evidences is one of the fundamental needs to data fusion, but this task is challenging in the decision-making domain because of the fusion of ever-increasing uncertain data. In this paper, a novel fuzzy-based multi-sensor data fusion method is proposed for fusing high-conflict uncertain data and avoiding the counter-intuition problem. Our key idea is to introduce the fuzzy inference mechanism into the similarity measurement model to measure conflict degree between the evidences. On this basis, belief entropy is used to calculate the uncertainty of evidences, so as to express the relative importance of the evidences. The reliability of evidences can be obtained by the credibility which is gained through the above method, and the quantitative information volume is used to revise each credibility degree to get the final weight according to the evidence. The numerical experimental results demonstrate that the presented method is feasible and effective in dealing with conflicting evidences. In addition, the application of fault diagnosis is given to show that the proposed approach is effective and advantageous compared with state-of-the-art approaches.