|
二维码（扫一下试试看！） |
| 废水处理系统的动态过程监测 |
| Dynamic Process Monitoring of Wastewater Treatment Systems |
| 收稿日期： |
| DOI：10.11980/j.issn.0254-508X.2019.02.009 |
| 关键词: 废水处理过程 故障检测 动态过程 动态主元分析 动态独立元分析 |
| Key Words:wastewater treatment processes fault detection dynamic process dynamic principal component analysis dynamic independent component analysis |
| 基金项目:南京林业大学大学生创新训练计划项目（2017NFUSPITP353）；制浆造纸工程国家重点实验室开放基金资助项目（201813）；南京林业大学高层次人才科研启动基金（163105996）。 |
|
| 摘要点击次数: 6929 |
| 全文下载次数: 1859 |
| 摘要:传统多元统计方法中，独立元分析（ICA）相较于主元分析（PCA）可有效提取信息的主要特征，保留更多原始数据。针对连续化生产所带来的动态特性，提出动态独立元分析（DICA）、动态主元分析（DPCA）,分别用来提升ICA和PCA的过程监测能力。结果表明，针对废水监测过程中偏移、漂移和完全失效3种传感器故障，DICA方法相较ICA的故障检测率在SPE统计量下分别提高了7.15%、18.58%和12.86%，故障检测率高达88.57%、84.29%及82.86%；DPCA故障检测在SPE统计量下相比于PCA也有一定提升，最高提高了28.57%，但其故障检测率要远低于DICA，这表明DICA方法对过程故障检测有较好的效果。 |
| Abstract:Fault detection is an important part in wastewater treatment processes. Independent component analysis(ICA）， as a method of multivariate statistical analysis， decomposed mixed information into linear combination of independent components， which can effectively extract the main information features of the process. Compared with principal components analysis(PCA）， ICA can extract more information from original date. In view of the dynamic characteristics of continuous production， dynamic independent component analysis(DICA） was proposed to improve the process monitoring ability of ICA. The results showed that the fault detection rates of DICA were optimized by 7.15%， 18.58%， and 12.86% for bias， drifting and complete faults in the wastewater data， respectively. The fault detection rates of DICA for the three kinds of faults were as high as 88.57%， 84.29%， and 82.86%， respectively. It indicates that DICA analysis method could significantly improve the process monitoring． |
| 查看全文 HTML 查看/发表评论 下载PDF阅读器 |