动态批次主元分析在化工过程中的应用研究(英文)

Application Research of Batch Dynamic Principal Component Analysis in Chemical Process

本文研究了批次动态主元分析(BDPCA)及其在化工过程中的应用。批次动态主元分析是主元分析的一种延伸,通过把三维矩阵转化成二维矩阵,结合时滞变量算法捕捉批次过程中的动态特性完成过程的监视并给出了确定时滞变量的算法.以TE过程为例,与多向主元分析法相比,仿真结果表明,BDPCA算法实现了考虑过程中的批次动态特性并提高了对过程变化的故障检测能力.

This paper presents the batch dynamic principal component analysis （BDPCA） and application in dynamic batch process. The method is an extension of the PCA. The three-dimensional matrix can be transformed to a bidimensional matrix and the lagged variables are used to capture the dynamic character- istics of the process. The algorithm of the time-lagged variable is proposed. BDPCA is applied to the well- known TE process. The results show that, compared with MPCA, the BDPCA has successfully simulated the batch dynamic process and enhanced the faults detection ability for the process variations.