Virtual linguistic trust degree-based evidential reasoning approach and its application to emergency response assessment of railway station

Abstract

In real life, the assessment information is usually expressed by virtual linguistic terms, but the trust degrees of different assessment values to the same reference grade are usually ignored. In addition, the provided information is usually incomplete or uncertain, because the objective things’ characteristics are elusive and the decision maker's knowledge is limited. To solve such problems, in this paper, we first propose the virtual linguistic trust degree. Then, in view of the nonlinear changes of the decision maker's psychological states, we depict the virtual linguistic trust degree by a nonlinear function, and present some relevant aggregation operators. In order to avoid the loss of information, we propose a novel evidential reasoning approach that combines the virtual linguistic trust degree with basic unit-interval monotonic function. In this approach, the normalized basic probability mass can be obtained by the continuous basic unit-interval monotonic function, which satisfies the consensus axiom of original evidential reasoning. Meanwhile, we propose the normalized factor on the basis of the rule that the remaining unassigned probability mass is assigned into any subsets, and then present the whole framework of an extended evidential reasoning algorithm. Furthermore, a numerical example about the emergency response assessment of railway station is conducted to show the usage of the algorithm. Finally, the validity of this algorithm is demonstrated by the comparative analysis.