According to phenomenon of missing data in the chemical process, a Locally Weighted Recovery Algorithm (LWRA) for dealing with missing data in the chemical process was proposed based on preserving the local data structure characteristic. The missing data points were located and marked with the symbol NaN (Not a Number), the missing data set was divided into complete data set and incomplete data set. The corresponding k nearest neighbors of incomplete data set were found in the complete data according to the size of integrity in turn, and the corresponding weights of k nearest neighbors were calculated according to the principle of minimum error sum of squares. Finally, the missing data points were reconstructed by k nearest neighbors and their corresponding weights. The algorithm was applied into two types of chemical process data with different missing rates and compared with two traditional data recovery algorithms, Expectation Maximization Principal Component Analysis (EM-PCA) and Mean Algorithm (MA). The results reveal that the proposed method has the lowest error, and the computation speed increases by 2 times in average than EM-PCA. The experimental results demonstrate that the proposed algorithm can not only recover data efficiently but also improve the utilization rate of the data, and it's suitable for nonlinear chemical process data recovery.

Multiway Principal Component Analysis (MPCA) has been widely used to monitor multivariate batch process. In MPCA method, the batch data are transformed as a vector in high-dimensional space, resulting in large computation, storage space and loss of important information inevitably. A new batch process fault diagnosis method based on the two-Dimensional Principal Component Analysis (2DPCA) was presented. Essentially, every batch data was presented as a second order vector, or a matrix. In this case, 2DPCA could be used to deal with the two-dimensional batch data matrix directly instead of performing vectorizing procedure with low memory and storage requirements. In addition, 2DPCA was used to model with the covariance average of all the batches, which accurately reflected the different faults and enhanced the accuracy of fault diagnosis to a certain extent. The monitoring results of an industrial example show that the 2DPCA method outperforms the conventional MPCA.