Extended accident scenario modeling based on Bayesian networks for risk evaluation
Xiaotao Li, Limin Tao, Mu Jia
Abstract
Xiaotao Li, Limin Tao, Mu Jia
Abstract
Conventional risk evaluation technique based on accident scenario such as event tree/fault tree suffer severe limitations of handling event dependencies and uncertainty. These dependencies and uncertainty are cumbersome to take into account when using standard event tree/fault tree modeling due to its clumsy structure and complicated quantitative solution. To make the accident scenario model more realistic, a method is proposed to explicitly represent the failures cascading effect dependency and uncertainty using Bayesian networks (BN). A simplified example of spacecraft hydrazine leak accident taken from literature illustrates the ideas presented above, and concludes that BN is a superior technique to fit a wide variety of accident scenarios profiting from its flexible structure and powerful reasoning.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Conventional risk evaluation technique based on accident scenario such as event tree/fault tree suffer severe limitations of handling event dependencies and uncertainty. These dependencies and uncertainty are cumbersome to take into account when using standard event tree/fault tree modeling due to its clumsy structure and complicated quantitative solution. To make the accident scenario model more realistic, a method is proposed to explicitly represent the failures cascading effect dependency and uncertainty using Bayesian networks (BN). A simplified example of spacecraft hydrazine leak accident taken from literature illustrates the ideas presented above, and concludes that BN is a superior technique to fit a wide variety of accident scenarios profiting from its flexible structure and powerful reasoning.
Key concepts: Fault tree analysis, Event tree, Computer science, Bayesian network, Event (particle physics), Accident (philosophy), Event tree analysis, Tree (set theory)