基于文本分析的故障序列模式挖掘算法  

Failures sequence pattern mining algorithm based on text analysis

作　　者：常文兵[1] 苑星龙 周晟瀚[1] 李磊 Chang Wenbing;Yuan Xinglong;Zhou Shenghan;Li Lei(School of Reliability & System Engineering,Beihang University,Beijing 100191 ,China)

机构地区：[1]北京航空航天大学可靠性与系统工程学院

出　　处：《计算机应用研究》2019年第9期2643-2646,2685,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(71501007);航空科学基金资助项目;北航研究生教育发展基金资助项目

摘　　要：针对结构化程度差、表达形式各异的文本数据,提出了一种基于文本信息的故障序列模式挖掘算法,用于发掘故障之间的时序关系。为从文本记录的故障信息中挖掘故障规律,首先将文本信息向量化,对故障文本信息进行相似度衡量,将表达相同意义的故障归为一类。在此基础上根据故障特性,提出最大窗口阈值、最小共现度阈值的概念,构建故障序列模式挖掘算法框架。最后对某型飞机文本故障信息进行序列模式挖掘,找出了正确的故障序列关系。实例验证了所提算法是正确有效的。For textual data with poor structured degree and different expression forms,this paper proposed a failures sequence pattern mining algorithm based on text information to explore the time sequence relationship between failures. In order to mine the failures rules from the text,firstly,it quantified the text information,measure the similarity of the failures information,and classified the failures that express the same meaning into one class.On this basis,it proposed the concept of maximum window threshold and minimum concurrence threshold based on failures characteristics,and built a mining algorithm framework for failures sequence pattern.Finally,it extracted sequential failures patterns from a certain aircraft,and found out the correct failures sequence relationship.The example shows that the proposed algorithm is correct and effective.

关 键 词：序列模型 数据挖掘 文本相似度 飞机故障 文本挖掘 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP301.6[自动化与计算机技术—计算机科学与技术]