警报
大洪水
计算机科学
数据挖掘
集合(抽象数据类型)
序列(生物学)
恒虚警率
领域(数学)
操作员(生物学)
人工智能
工程类
地理
数学
抑制因子
遗传学
航空航天工程
生物化学
生物
考古
纯数学
化学
基因
程序设计语言
转录因子
作者
Matthieu Lucke,Moncef Chioua,Chriss Grimholt,Martin Hollender,Nina F. Thornhill
标识
DOI:10.1016/j.jprocont.2019.04.010
摘要
a b s t r a c tDuring an alarm flood, the alarm rate is greater than the operator can effectively manage.Many alarm data analysis methods have been proposed in the literature to mitigate the impact of alarm floods.This paper gives a review of the state of the art in alarm data analysis and aims at structuring the field.A distinction between sequence mining methods that apply to alarm sequences and time series analysis methods that apply to alarm series is suggested.The review highlights that online applications to help the operators during alarm flood episodes have been only treated as sequence mining problem in the literature to date.To address this gap, the paper also presents a binary series approach to classify ongoing alarm floods based on a set of historical alarm floods.The motivation for a binary series approach is demonstrated through an industrial case study of a gas-oil separation plant, and the performance of the presented method is compared with the performance of an established sequence alignment method.
科研通智能强力驱动
Strongly Powered by AbleSci AI