Dependence assessment in human reliability analysis based on D numbers and AHP

Since human errors always cause heavy loss especially in nuclear engineering, human reliability analysis (HRA) has attracted more and more attention. Dependence assessment plays a vital role in HRA, measuring the dependence degree of human errors. Many researches have been done while still have improvement space. In this paper, a dependence assessment model based on D numbers and analytic hierarchy process (AHP) is proposed. Firstly, identify the factors used to measure the dependence level of two human operations. Besides, in terms of the suggested dependence level, determine and quantify the anchor points for each factor. Secondly, D numbers and AHP are adopted in model. Experts evaluate the dependence level of human operations for each factor. Then, the evaluation results are presented as D numbers and fused by D number's combination rule that can obtain the dependence probability of human operations for each factor. The weights of factors can be determined by AHP. Thirdly, based on the dependence probability for each factor and its corresponding weight, the dependence probability of two human operations and its confidence can be obtained. The proposed method can well address the fuzziness and subjectivity in linguistic assessment. The proposed method is well applicable to assess the dependence degree of human errors in HRA which inherently has a linguistic assessment process.