引用本文:常鹏,乔俊飞,张祥宇,王普.基于四阶累积分析的工业大肠杆菌制备过程故障诊断[J].控制理论与应用,2020,37(3):667~675.[点击复制]
CHANG Peng,QIAO Jun-fei,ZHANG Xiang-yu,WANG Pu.Fourth order cumulant analysis based fault diagnosis of the preparation process of industrial Escherichia coli[J].Control Theory and Technology,2020,37(3):667~675.[点击复制]
基于四阶累积分析的工业大肠杆菌制备过程故障诊断
Fourth order cumulant analysis based fault diagnosis of the preparation process of industrial Escherichia coli
摘要点击 1830  全文点击 640  投稿时间：2018-05-17  修订日期：2019-05-20
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2019.80361
  2020,37(3):667-675
中文关键词  多向核熵独立成分分析  四阶累积分析  多向核主成分分析  多向核独立成分分析
英文关键词  multi-way kernel entropy independent component analysis  forth-order cumulant analysis  multi-way kernel principal component analysis  multi-way kernel independent component analysis
基金项目  国家自然科学基金
作者单位E-mail
常鹏* 北京工业大学 changpeng@bjut.edu.cn 
乔俊飞 北京工业大学  
张祥宇 北京工业大学  
王普 北京工业大学  
中文摘要
      针对工业大肠杆菌制备过程具有的非线性和非高斯性共存的故障诊断难于有效对故障源进行定位的问题，提出基于多向核熵独立元分析(Multiway Kernel Entropy Independent Analysis-MKEICA)的过程监测方法。该方法首先将MKEICA方法作为工业实际生产过程监测模型；其次针对传统监控统计量T2、I2和SPE为低阶监控统计量的不足提出四阶累积监控统计量用于工业过程监测；再次针对四阶累积监控统计量推导其故障产生的原因；最后将其应用在实际的工业过程并与多向核独立元分析(Multiway Kernel Independent Component Anlaysis-MKICA)监测模型进行对比验证该方法的高效性，同时对故障变量进行追溯，验证该方法的实用性。
英文摘要
      Aiming at the problem of nonlinear and non-Gaussian coexistence fault diagnosis in the process of industrial Escherichia coli preparation, it is difficult to locate the fault source effectively. A process monitoring method based on MKEICA is proposed. Firstly, the MKEICA method for the industrial production process monitoring model; secondly, the traditional T2, I2 and SPE monitoring statistics for low order monitoring statistics proposed fourth order statistics monitoring statistics for industrial process monitoring; once again for the three order cumulative monitoring statistics derived the fault generated; the method can be applied in real industrial processes and the effectiveness of the method is verified and compared the conventional MKICA monitoring model the fault trace, verify the practicability of this method.