中文摘要：

针对多向主元分析（multi—wayprincipalcomponentanalysis，MPCA）算法用于间歇过程实施监控时需要将三维数据转换为高阶的二维矩阵，从而易导致算法的计算量大，且会丢失一些有用信息的情况进行了研究，提出了一种新的间歇过程故障诊断方法——二维主成分分析法（2一dimensionalprincipalcomponentanalysis，2DPCA）。该算法首先利用各个批次的二维矩阵构造协方差矩阵，进而求得所有批次协方差矩阵的平均值进行建模，大大降低了计算复杂度，运算时间较MPCA缩短了19／20到3／4，且无须占用太多存储空间；同时，2DPCA计算协方差矩阵较MPCA更为准确，取协方差矩阵的平均值能够更加精确地反映不同类型的故障，在一定程度上增强了故障诊断的准确率。最后，通过将所提出的方法应用于青霉素发酵过程的监控中，验证了该算法的有效性和准确性。

英文摘要：

The three-dimensional data for batch process monitoring needed to transform as a vector in high-dimensional space using MPCA algorithm, resulting in the large amount of computation and the loss of some useful information. This paper pro- posed the new method for batch process fault diagnosis, which based on the two-dimensional principal component analysis （2DPCA）. The method took advantage of the two-dimensional matrix in each batch to construct the eovariance matrix, and ob- tained the average of the covariance matrix of all batches to model,which leaded to reduce the computational complexity, short the operation time from 19/20 to 3/4 compared with MPCA and took up the little storage space. At the same time, the 2DPCA approach was more accurate than MPCA when calculating the covariance matrix. 2DPCA was used to model with the eovariance average of all batches, which accurately reflected the different faults and enhanced the accuracy of fault diagnosis. Finally, the proposed method was used in the penicillin fermentation process. The results demonstrate the validity and accuracy of the algo- rithm.