摘要
针对TEP(Tennessee Eastman Process)的故障诊断问题,分别采用PCA主元分析法与粒计算属性约简算法对TEP的52个变量在15种故障情况下的实测数据分别进行处理。结果显示采用PCA主元分析方法可以将其中的14个对各种故障情况都无影响或影响微弱的变量排除,而采用基于粒计算的属性约简算法。对预处理后的数据进行属性约简,可以将条件属性约简至24个即排除28个条件属性,表明该方法对TEP故障诊断的有效性。
In light of TEP（ Tennessee Eastman Process） fault diagnosis issue, we use the principal component analysis （PCA） and granular computing attribute reduction algorithm respectively to process the measured data of 52 variables in TEP under 15 kinds of fault conditions separately. Results show that by applying PCA, 14 variables among them can be excluded of which the various fault conditions have either no or slight influence on them, while using the granular computing-based attribute reduction algorithm to carry out attribute reduction on the data with pretreatment, it can reduce the conditional attributes to 24, i. e. , 28 conditional attributes are excluded. This demonstrates the effectiveness of the method on TEP fault diagnosis.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第7期82-85,共4页
Computer Applications and Software
关键词
粒计算
主元分析
属性约简
TEP
Granular computing
Principal component analysis
Attribute reduction
TEP
