中文摘要：

针对现有告警管理方法无法判断炼化装置运行中的过渡过程,对过渡过程不能进行准确有效地监测和管理,提出了基于贝叶斯估计的动态告警线计算方法。通过训练历史过渡过程数据得到先验概率,自适应判断过渡过程和估计动态告警线,解决了传统告警线的斜率不能赋值且只能依靠人为调节的问题,并可通过判断系统状态调节告警管理方式。经常压塔原油进料流量调整和减压炉干气流量工艺调节的过渡过程现场数据验证,结果表明：与传统告警管理方法相比,基于贝叶斯估计的动态告警管理方法的误告警总数量减少了87.34%,避免了告警洪水的发生,提高了炼化工艺运行的安全性和可靠性。

英文摘要：

Aiming at the problems that the existing alarm management methods can’t judge the transient processes when the refining device is operating,and they can’t monitor and manage the transient processes accurately and effectively,a calculation method of dynamic alarm line based on Bayesian estimation was proposed. The prior probability was obtained by training the historical data of transient processes,then the transient processes were adaptively judged,and the dynamic alarm line was estimated. The problem that the slope of traditional alarm line can ’t be assigned and can only be adjusted manually is resolved,and the alarm management mode can be adjusted through judging the system status. Through field data verification of transient processes including flow adjustment on crude oil feed in atmospheric tower and technical adjustment on dry gas flow in vacuum furnace,it showed that compared with traditional alarm management methods,the total amount of error alarm by the dynamic alarm management method based on Bayesian estimation reduced by 87. 34 %,which avoided the occurrence of alarm flooding..