石化装置火灾爆炸贝叶斯网络与防护层集成分析  

Integrative analysis of Bayesian networks and LOPA for fire and explosion in petrochemical plant

作　　者：姚锡文[1] 许开立[1] 王贝贝[1] 闫放[1] 王延瞳 王文菁[1] 

机构地区：[1]东北大学资源与土木工程学院,辽宁沈阳110819

出　　处：《中国安全生产科学技术》2014年第3期95-100,共6页Journal of Safety Science and Technology

基　　金：辽宁省自然科学基金资助项目(2013020137)

摘　　要：复杂的石油化工装置在运转过程中存在诸多不确定因素,易发生火灾、爆炸等重大事故,给安全生产带来极大威胁。考虑到传统的系统安全分析方法在风险评估中存在一定局限性,引入贝叶斯网络与防护层集成分析模型。应用GeNIe软件将系统故障树转成贝叶斯网络,根据贝叶斯双向推理进行故障预测和诊断,快速识别系统薄弱环节并确定为风险贝叶斯故障节点,结合防护层分析提出相应的独立防护层,确定剩余风险水平。实例应用表明,所构建的贝叶斯网络与防护层集成分析模型对复杂系统进行风险评估是可行的,较传统的事件树、故障树分析方法更加科学、合理。There are many uncertain factors in the running complex petrochemical plants, it is prone to major acci- dent such as fire, explosion, and brings about huge threat in safety production. Considering the limitations of tradi- tional system safety analysis methods, an integrative analysis model of Bayesian networks combining LOPA was pro- posed. The system fault tree was turned to Bayesian networks by GeNie software, then the fault prediction and diag- nosis were conducted by two-side reason function of Bayesian networks, the system weakness was rapidly identified, to be Bayesian risk fault node, then the corresponding independent protection protective layer was proposed and was determined combining with LOPA, so the residual risk level was determined. The case application showed that the integrative analysis model of Bayesian networks combining with LOPA is practicable for complex system risk analy- sis, and it is more scientific and reasonable than the traditional event tree analysis and fault tree analysis.

关 键 词：石化装置 火灾 爆炸 贝叶斯网络 风险 防护层分析 

分 类 号：X937[环境科学与工程—安全科学]