摘要
为提高非常规突发灾害事故情景分析的有效性和全面性,在界定非常规突发灾害事故情景概念的基础上,基于知识元表示非常规突发灾害事故的情景,利用动态贝叶斯网络对非常规突发灾害事故情景演变进行构建;以大连"7·16"中石油储运公司油库火灾为例,演示基于知识元和动态贝叶斯网络的非常规突发灾害事故情景分析方法的具体流程,并对推演结果进行分析。推演结果显示,18时12分输油管线爆炸起火的概率为72.0%,18时19分T103罐爆炸起火的概率为92.6%,21时30分泵房爆炸起火的概率为87.5%,与实际情况基本一致,证明了该方法的有效性和可行性。
This article intends to provide a scenario analysis of un conventional emergency disaster accidents based on the knowledge el ements and dynamic Bayesian networks. The socalled knowledge el ements we mean to say may include ＂the Case, the Disaster, the Ob ject, the Activity＂ when they are used to represent the scenario of some unconventional emergency disaster accidents. It is on the basis of introducing the related theories of Dynamic Bayesian Networks that we resort to all the network node variables and set thresholds. On the condition that the node variables are greater than the threshold value, they should be considered to be the key knowledge elements that tend to ＇affect the catastrophe, and vice versa. However, if the key knowl edge element that influences the said disaster, it is likely to be ap plied to build up the evolution network of the scenarios of unconven tional emergency disasters, it would also be possible to be used to work out the insitu probability of each node variable by utilizing the joint probability formula so as to achieve a scenario deduction. The present paper also wants to quote the ＂July16＂ Dalian oil depot fire as a case study sample to analyze the deduction in hoping to demon strate the specific process of the scenario evolution analysis model. The result of our analysis shows that： the probability of the pipeline bursting and blowout is about 72% at 18：12, whereas the possibility of the T103 tank bursting and blowout symbolized a 96% at 18：19, and, in the same way, the pump bursting and blowout is 85% at 21 ： 30. The result is just in correspondence with the actual scenario in situ, which proves the effectiveness and feasibility of this method. It is just because of the complex, diverse and uncertain characteristic features of the influencing factors of unconventional emergency disas ter accident, the key information elements may symbolize the higher demands on the sample information. Therefore, it is only when the sample information and data is enough representative that the suffi ciently experienced experts in the field can predict the probability of the accidents more accurately through logic deduction.
出处
《安全与环境学报》
CAS
CSCD
北大核心
2014年第4期165-170,共6页
Journal of Safety and Environment
关键词
安全工程
非常规突发灾害事故
知识元
动态贝叶斯网络
应急指挥决策
safety engineering
unconventional emergency disaster accident
knowledge element
dynamic Bayesian net- works
decision of emergency command
