Abstract:

In order to find out root alarms so as to avoid alarm flooding in industrial processes, this paper suggests a fuzzy weighted association rules mining method which integrates fuzzy sets, the Apriori mining algorithm and time-series analysis to extract the association rules from alarm sequence data. Considering the temporal characteristics and similarity attributes of the alarm data, the similarity degrees are assigned as the weights of the association rules, enhancing the algorithmic efficiency and accuracy. Further, in contrast to quantitative knowledge representations, the resultant fuzzy association rules are easier identified to the operators. Practical data of an industrial process are employed to illustrate the benefits of the proposed method.

Key words: consequential alarms, data similarity, causality, association rules

摘要：

面向寻找工业报警序列根源,抑制报警泛滥,论文提出了一种模糊加权关联规则挖掘方法,结合模糊集合、Apriori数据挖掘算法和时间序列分析挖掘报警关联规则。基于报警数据的时间约束属性和相似度约束属性,利用相似度作为模糊加权关联规则挖掘算法的权重,提高挖掘效率和准确性。并且,相对于定量表达,模糊关联规则对于操作者来说更加容易使用。工业实例验证了方法的有效性。

关键词: 相关报警, 相似数据, 因果关系, 关联规则