论文部分内容阅读
针对间歇过程独特的数据特点,提出1种将因子分析(FA)作为独立成分分析(ICA)白化预处理手段的多向因子分析白化独立成分分析(multiway factoranalysis-independent component analysis,MFA-ICA)间歇过程监控方法。因子分析充分考虑了模型误差的普遍意义,拥有优秀的噪声建模能力。将其代替主成分分析用于白化,可以更好的提取数据集的本质信息。首先将间歇过程三维数据依次按批次和变量展开得到二维数据矩阵,接着把上述方法用于展开后的数据,利用ICA的I~2统计图实现在线故障检测。该方法用于标准仿真平台Pensim,结果表明上述方法对于提高间歇过程故障检测的快速性,降低漏报率有明显效果。
Aiming at the unique data characteristics of intermittent process, a multi-factorial analysis of factorial analysis (FA) as an independent component analysis (ICA) pretreatment method was proposed. Multiway factoranalysis-independent component analysis (MFA-ICA) Process monitoring methods. Factor analysis takes full account of the general significance of model errors and has excellent noise modeling capabilities. Use it instead of principal component analysis for whitening, which can better extract the essential information of the data set. First of all, the three-dimensional data of batch process is expanded by batch and variable to obtain a two-dimensional data matrix, and then the above method is used for the expanded data, and the on-line fault detection is realized by using I ~ 2 charts of ICA. The method is applied to the Pensim standard simulation platform. The results show that the above method has obvious effect on improving the quickness of fault detection and reducing the false negative rate in batch process.