基于贝叶斯模型的化工过程动态风险研究

发布时间：2018-06-13 11:05

本文选题：贝叶斯理论 + copula函数　；参考：《华东理工大学》2015年硕士论文

【摘要】：随着化学工业的快速发展,化工对象生产负荷不断提高,其安全问题受到越来越高的重视。目前,化工过程风险分析领域的相关研究主要集中于如何有效实现对化工过程的故障检测和诊断,而很少有专家从化工过程安全传导机理出发建立化工过程事件的评价模型。为此,本文基于贝叶斯模型建立化工过程班组操作的动态风险评估模型。本文以特定的化工生产装置为例,利用基于事故序列先导数据的贝叶斯模型来定量分析班组人员操作对化工风险的影响。轮班班组的操作能力以及不同班组之间的关联性对化工生产过程的安全性影响很大,而copula函数可以体现不同变量之间复杂的非线性关系。另外,copula函数种类繁多,然而在数据很少的情况下copula的选取至关重要。为了排除人为给定先验参数的影响,在此利用无信息先验的最大熵方法选择copula函数类型。因此,考虑到不同班组操作的时序性和耦合性,本文利用贝叶斯理论结合copula函数,提出了基于班组操作机理分析的最大熵条件概率模型和基于班组操作数据信息的MCMC抽样分析方法来定量地评估班组在事件树模型中操作能力的风险差异性。此外,在异常事件发生之后,系统运行的最终状态受到班组操作的影响,通过研究每个时间段内班组操作导致3类运行状态发生的概率来定量评估不同班组操作的可靠性。因此,在建立高温异常事件树模型的基础上,本文通过分析和比较每个班组操作期间所导致装置出现不同运行状态的风险概率均值来反映班组整体的操作能力。
[Abstract]:With the rapid development of chemical industry, the production load of chemical industry is increasing, and the security problem has been paid more and more attention. At present, the related research in the field of chemical process risk analysis is mainly focused on how to effectively realize the fault detection and diagnosis of chemical process, but few experts start from the mechanism of chemical process safety transmission. For this purpose, this paper establishes a dynamic risk assessment model for chemical process team operation based on Bayes model.
This paper takes a specific chemical production device as an example to analyze the effect of the group operation on the chemical risk by using the Bayesian model based on the pilot data of the accident sequence. The operation ability of the shift group and the association between the different groups have a great influence on the safety of the chemical production process, and the copula function can reflect the difference. The complex nonlinear relationship between variables. In addition, there are a wide variety of Copula Functions. However, the selection of Copula is very important in the case of little data. In order to exclude the influence of human given prior parameters, the maximum entropy method without information prior is used to select the copula function type. By combining the Bayesian theory with the copula function, the maximum entropy conditional probability model based on the analysis of group operation mechanism and the MCMC sampling analysis method based on the group operation data information are proposed to quantitatively evaluate the risk difference of the operation ability of the group in the event tree model.
In addition, after the occurrence of abnormal events, the final state of the system is affected by the operation of the group. By studying the probability of the 3 classes of running states in each time period, the reliability of the operation is evaluated quantitatively. Therefore, on the basis of the establishment of a high temperature anomaly event tree model, this paper analyzes and compares the results. The average risk probability of different operation states caused by each team operation reflects the overall operational ability of the team.
【学位授予单位】：华东理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TQ086;TP18

【参考文献】