详细信息

基于峰度的非线性独立元分析在典型化工过程故障诊断中的应用  ( EI收录)  

Industrial Process Monitoring and Fault Diagnosis with Nonlinear Independent Component Analysis Based on Kurtosis

文献类型:期刊文献

中文题名:基于峰度的非线性独立元分析在典型化工过程故障诊断中的应用

英文题名:Industrial Process Monitoring and Fault Diagnosis with Nonlinear Independent Component Analysis Based on Kurtosis

作者:吴友丰[1];王振雷[1];钱锋[2]

机构:[1]华东理工大学信息科学与工程学院,上海200237;[2]华东理工大学化学工程联合国家重点实验室,上海200237

年份:2008

卷号:34

期号:4

起止页码:568

中文期刊名:华东理工大学学报(自然科学版)

外文期刊名:Journal of East China University of Science and Technology

收录:CSTPCD;;EI(收录号:20083911603599);Scopus;北大核心:【北大核心2004】;CSCD:【CSCD2011_2012】;

基金:国家杰出青年科学基金(60625302);国家自然科学基金(60704028);上海市基础研究重点项目(07JC14016);长江学者和创新团队发展计划(IRT0721);高等学校学科创新引智计划(B08021);上海市重点学科建设项目(B504)等资助

语种:中文

中文关键词:非线性独立元分析;峰度;故障诊断;TE过程

外文关键词:nonlinear independent component analysis; kurtosis; fault detection; Tennessee Eastman process

摘要:针对实际工业过程数据中的非线性和冗余度问题,提出了一种基于峰度的非线性独立元分析方法。该方法根据互信息最小化原理提取独立分量,按峰度进行降维,去掉冗余数据,尽可能地保留了反应工业过程变化本质特征的独立分量。将该方法应用于TE过程监控,并与传统的线性独立元方法相比较,结果表明:该方法具有较高的灵敏度和较低的漏报率。
Considering the nonlinear and redundant characteristic of the data in real industry process, a nonlinear independent component analysis based on kurtosis is presented. This method extracts independent components according to mutual information minimize theory and reduces the dimension in term of the kurtosis in order to remove redundant data and reserves the independent component which represent the state of industrial process. The NICA is applied to the fault detection of the Tennessee Eastman process and compared with the linear ICA. The simulation results show that the NICA has better performance in sensitivity and lower underreport rates.

参考文献:

正在载入数据...

版权所有©华东理工大学 重庆维普资讯有限公司 渝B2-20050021-7 
渝公网安备 50019002500408号 违法和不良信息举报中心