基于过程数据的关联报警分组和抑制策略

A grouping and suppression strategy for correlated alarms based on process data

针对目前流程工业报警系统中普遍存在的报警系统效率低、连锁报警多的现象,提出一种基于过程数据的关联报警分组和抑制策略。该策略采用离差平方和法对过程数据进行聚类分析,根据距离系数把关联的报警变量分组;并利用报警优先级量化规则,从响应时间和后果严重度两方面计算报警的优先级分值,以此确定报警小组中需要及时响应和抑制的报警。该策略有助于控制员理解报警间的关联关系、提高对连锁报警的响应效率。最后在TE模型中运行测试,验证了该策略的有效性和实用性。

To solve the common problems of the alarm system in the process industry,such as low efficiency and frequent interlocking alarm,this paper proposes a grouping and suppression strategy for correlated alarms based on process data.The Ward Method was used in this strategy for cluster analysis of the process data,and distance coefficient was the consideration to group the alarm variables that may related.Quantization regulations of alarm priority were introduced to identify the alarms needing response or suppression in an alarm group via analyzing the accurate priority score from both response time and severity of consequences.It is helpful for controllers to recognize the correlation among alarms,and improve the response efficiency.The strategy yields an effective alarm management that can help plant owners and operators to comply with the standards for alarm management such as ANSI/ISA 18.2（2009）and EEMUA 191（2007）which set limits on the number of alarms per unit time for an operator.The effectiveness of the approach is illustrated by successful application in TE process model where a significant reduction of alarms has been achieved.