Dependence assessment in human reliability analysis using the 2-tuple linguistic information and DEMATEL method

Abstract

Human reliability analysis (HRA), which is to analyze human contribution to system risk, is an effective way to model and assess human errors. Dependence assessment among human errors is an essential part of HRA, which often depends on the judgments of experts. As real-world problems often involve many complex factors, the judgments provided by the experts are often linguistic terms, even under uncertainty. To this end, by integrating 2-tuple linguistic variables and DEMATEL method, this paper presents a novel way to assess the dependence among human actions in HRA. In the proposed method, the linguistic judgments of the experts are modeled using 2-tuple linguistic variables, and the weights of the influential factors are determined using DEMATEL method, furthermore, the conditional human error probability is calculated by aggregating the 2-tuples of different influential factors based on the 2-tuple weighted average operator, where a novel weight calculation method is developed to determine the weights of different experts. Finally, a case study is presented to demonstrate the effectiveness and reliability of the proposed method. By adopting 2-tuple linguistic variables and DEMATEL method, the proposed method could effectively address the uncertainty in the dependence assessment while capturing the relationship among the influential factors.