摘要
现阶段,“报警泛滥”现象普遍存在,其主要由于报警阈值设置不合理导致,这种现象超出操作员操作范围。化工过程关联特征复杂,异常工况发生易诱发多个变量报警,融合专家知识和数据统计的方法挖掘报警变量间的内在关联,结合优先级进行多变量报警阈值优化研究。采用主元分析法初步确定变量的重要程度,针对数据统计的缺陷,通过标定风险对变量进行优先级划分修正权重,进而结合标准EEMUA 191—2013分配报警数目、优化报警阈值。通过TE模型验证,此方法能够有效减少17%~46%的报警数量。
Due to the unreasonable setting of alarm threshold,there are still a large number of problems of alarm flooding.In order to avoid the interference of too many invalid alarms and reduce the number of alarms,the alarm threshold is optimized based on knowledge and statistics.Firstly,the correlation between variables was analyzed by PCA(principal component analysis),and the importance of variables was determined,then the threshold was obtained by dividing the alarm number according to the weight and the international standard.However,the obtained threshold only relies on the data,which is one-sided.Therefore,on this basis,the variables are prioritized by combining the parameters such as consequence,time,frequency and so on,the weight is corrected,and the alarm number is allocated and the alarm threshold is optimized in accordance with the international standard.Verified by TE model,this method can effectively reduce the number of alarms by 17%~46%.
作者
杨振巍
王诚儒
雷方怡
Yang Zhenwei;Wang Chengru;Lei Fangyi(Safety and Environmental Protection Division,Jinxi Petrochemical Branch,Huludao 125000,China;College of Chemistry and Chemical Engineering,China University of Petroleum(East China),Qingdao 266580,China)
出处
《山东化工》
CAS
2023年第19期212-215,共4页
Shandong Chemical Industry
关键词
报警阈值
多变量
主元分析
优先级
alarm threshold
multi-variable
principal component analysis
optimization
safety
