Dependence Assessment in Human Reliability Analysis Using Evidence Theory and AHP

Area of Science:

• Engineering
• Risk Analysis
• Cognitive Science

Background:

• Dependence assessment in Human Reliability Analysis (HRA) is crucial but often relies on subjective expert opinions, leading to inconsistencies.
• Existing methods for evaluating the dependence among human errors lack objectivity and can be difficult to standardize.

Purpose of the Study:

• To propose a novel computational model for assessing dependence among human errors in HRA.
• To enhance objectivity and consistency in human error dependence evaluation.
• To address ambiguity and confidence levels in expert judgments within HRA.

Main Methods:

• Integration of the Dempster-Shafer evidence theory (DSET) with the analytic hierarchy process (AHP) method.
• Identification of dependence influencing factors and expert-driven weighting using AHP.
• Representation of analyst judgments as basic belief assignments (BBAs) and fusion using DSET.

Main Results:

• The proposed model successfully integrates expert opinions and analyst judgments into a fused BBA.
• The model quantines the Common Human Error Probability (CHEP) based on the fused BBA.
• Demonstrated ability to handle ambiguity and varying degrees of confidence in judgments.

Conclusions:

• The DSET-AHP model offers a more objective and consistent approach to dependence assessment in HRA.
• This computational method mitigates subjectivity inherent in traditional expert-based evaluations.
• The model provides a framework for improved human reliability assessment in complex systems.