Toward Efficient Process Monitoring Using Spatiotemporal PCA

Abstract

Principal component analysis (PCA) has shown its high efficiency in process monitoring (PM). However, most of the existing PCA-based PM approaches only consider the spatial prior and ignore the temporal prior. Therefore, in this brief, we propose a novel PM framework using spatiotemporal PCA (STPCA), which incorporates both spatial and temporal priors. Technically, the spatial prior is integrated to preserve the cause-effect relationship of process variables, and the temporal prior is embedded to maintain the geometric structure of process samples. Moreover, an efficient optimization algorithm is developed using the alternating direction method of multipliers (ADMM) in a symmetric Gauss-Seidel (sGS) manner. Finally, the improved monitoring performance is verified on the benchmark Tennessee Eastman (TE) process. In particular, compared with PCA, the fault detection rate of fault IDV(20) is increased by 9.88%. This suggests that the proposed framework is promising for PM.