基于贝叶斯理论与Vine Copula的化工过程异常事件数的预测

Forecasting Abnormal Event Numbers in Chemical Process with Bayesian Theory and Vine Copula

针对化工过程风险,提出了一种化工过程异常事件数的预测方法。化工生产过程中由于受到干扰,时常发生异常事件。异常事件如果得不到有效控制将引发生产事故,其发生次数越高表明发生生产事故的概率越大,因此,准确预测化工过程异常事件数有助于提高化工过程的风险管理水平。基于操作班组,采用贝叶斯理论与Vine Copula建立了动态预测模型,实现对化工过程一个轮班内异常事件数的预测。

A prediction method is proposed to cope with abnormal event numbers, which often appear in chemical process due to external disturbance. If an abnormal event is not effectively controlled, it will probably result in an accident. The higher abnormal event numbers are, the greater the probability of production accidents is. Therefore, the precise prediction on the abnormal event numbers can effectively improve the risk management level on the chemical process. Usually, four operating teams work in a workshop and the number of abnormal event varies from team to team. Based on operating teams, the dynamic prediction model is constructed by using the Bayesian theory and Vine Copula such that the abnormal event numbers in a operating team can be predicted.