The combined use of evidence theory and vague sets to interpret multimodal inspection data

Several measurement modalities have been applied for the safety and reliability evaluation of complex system. A great deal of information can be obtained by multimodal inspection. But different sensors can only capture part of the exterior and interior geometry since the limitations of its involved physical phenomena. So a challenging problem is how to effectively interpret the available multimodal inspection data, especially when the data show vague, uncertain and even conflict information. In this paper, a combined vague sets/ D-S evidence theory approach is proposed to make more reasonable inferences using multi-source information fusion. Since the D-S theory shares the similar form in the fundamental definition of the measure of a proposition to that in the definition of grade membership of an element in vague sets, the relationship between the true-membership function and false-membership function of vague sets and belief/plausibility functions of D-S theory is discussed. Based on the feature of vague sets, the true-membership function and false-membership function are used to describe the belief level of fusion target. Then they are combined by an enhanced combination algorithm based on D-S evidence conventional combination rule. Finally, an example is conducted to demonstrate the effectiveness of the proposed combined vague sets/ D-S evidence theory. According to its firm mathematical foundation, the proposed approach can express and handle uncertain and vague information effectively, and can be applied to fuse any bodies of multimodal inspection data without changing the recursive combination algorithm.