Risk analysis of chemical accidents based on data-driven Bayesian network
To reduce the subjective intervention in risk analysis of chemical plants,a data-driven risk analysis model based on the association rules and Bayesian network was constructed,which covered three task items,including data set item,associa-tion rule driven item and Bayesian network risk assessment item.Firstly,the accident reports and accident factors were col-lected to construct an accident database.Secondly,the accident data were imported into the Apriori algorithm,and the struc-ture of Bayesian network and conditional probability table were determined based on the factor correlation of association rules.Then,the prior probability and conditional probability were calculated according to the occurrence frequency of accident fac-tors,and the sensitivity of factors was calculated by Fussel-Vesely.Finally,a chemical poisoning and asphyxiation accident was used as an example,and the influence of the accident factors was assessed using the data-driven risk analysis model.The research results can provide new ideas to reduce and avoid chemical accidents,and help improve the overall safety level of re-lated enterprises.
万方数据
维普
