Quantitative analysis methodology of non-deterministic causal relationship in risk analysis

Abstract

In the safety-critical systems, potential hazard may jeopardize the safety of system, which can even lead to catastrophic accident. The occurrence of accident contains deterministic and non-deterministic causal logic relationships. The former refers to technical aspects, where event tree and fault tree could be applied to analyze. The latter mainly refers to operational aspects. Due to uncertainties and lack of data, Bayesian network becomes the appropriate tool. However, how to determine the conditional probability table is a difficult task. In this paper, a new conditional probability allocation method based on fuzzy logic is proposed to reference posterior probability. Before quantitative analysis, the risk model of influencing factors may be established. Finally, the proposed method is applied to evaluate human error probability.