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 studies batch dynamic principal component analysis (BDPCA) and its application in the chemical process. Batch dynamic principal component analysis is an extension of principal component analysis. By transforming the three-dimensional matrix into a two-dimensional matrix, and combining the time-varying variable algorithm to capture the process of dynamic characteristics during the batch process, Taking the TE process as an example, compared with the multi-way principal component analysis, the simulation results show that the BDPCA algorithm can realize batch dynamic characteristics and improve the ability of fault detection of process changes.