Chemical process risk dynamic analysis method based on Bayes and accident tree

Abstract

The invention discloses a Bayesian dynamic analysis method established based on the Bayesian theory, which belongs to the field of safety system risk analysis and is suitable for dynamically updatingthe failure rate of a safety system and the risk of variables in real time. The method comprises the steps of 1) establishing a Bayesian correction model in combination with accumulated data and the Bayesian theory; 2) analyzing a result event probability according to variable probability posteriori distribution and variable probability sensitivity distribution; and 3) for the key variable with arelatively high probability of triggering the HH / LL alarm by the H / L alarm, enabling an engineer to improve a control link of the key variable and provide a reference for an operator of a factory. According to the invention, small accidents such as high-low alarm are used as result events, and are easier to collect, analyze and apply in practice; the dynamic analysis of the event is analyzed in combination with an event tree model to obtain an updated probability value of a result event; and the problem that a traditional analysis method cannot dynamically update the risk probability is solved.

Description

technical field [0001] The present invention establishes a Bayesian dynamic analysis method based on Bayesian theory, belongs to the field of safety system risk analysis, is suitable for real-time dynamic updating of safety system failure rate and variable risk, and is a "near miss" based on near accident alarm data. Bayesian + event tree" dynamic analysis method of chemical process risk. Background technique [0002] Traditional risk analysis methods only consider major accidents such as fire, explosion or poisoning from a static point of view to evaluate risks, often ignoring small accidents and potential accidents. Since major chemical accidents are low-frequency events, the limited data makes it difficult to rely on traditional chemical accident data. Risk assessment methods cannot be effectively applied to production practice. At the same time, compared with the final chemical accidents, the high frequency and high data volume of a large number of near misses can provid...

AI Technical Summary

Problems solved by technology

[0002] Traditional risk analysis methods only consider major accidents such as fire, explosion or poisoning from a static point of view to evaluate risks, often ignoring small accidents and potential accidents. Since major chemical accidents are low-frequency events, the limited data makes it difficult to rely on traditional chemical accident data. Risk assessment methods cannot be effectively applied to production practice. At the same time, compared with the final chemical accidents, the high frequency and high data volume of a large number of near misses can provide another way of thinking for chemical process risk assessment.

Images