Fuzzy Risk Assessment Based on Interval Numbers and Assessment Distributions

Abstract

In risk assessment problems, multiple experts are often involved. On many occasions, assessment experts cannot give crisp scores even if there are scoring standards because of limitation and uncertainty. It is reasonable that the assessment data are expressed in the form of interval numbers with self-confidence. A fuzzy risk assessment model based on interval numbers with self-confidence is proposed in this paper. First, a multi-expert interval numbers with self-confidence fusion model is constructed. In the model, experts weights are determined based on subjective weights and objective weights, and the objective weights are calculated based on the length of the base of interval number and self-confidence simultaneously. Second, a novel method determining the symbolic proportion in an assessment distribution is proposed. This method integrates the concept of similarity measure between generalized fuzzy numbers and the length of the base of intersection of fuzzy numbers. Some properties of the proposed symbolic proportion measure are proved. Third, a fuzzy risk assessment model is proposed based on the fuzzy inference system. The output of the model is a distribution with the possible risk ranks and corresponding symbolic proportions. Finally, an illustrative example which shows the proposed fuzzy risk assessment model effective is demonstrated.