Multivariate Statistical Process Monitoring Based on Synchronization of Batch Trajectories Using DTW

Abstract An improved Dynamic Time Warping (DTW) algorithm is presented which can be used in on-line fault detection. The two new monitoring approaches, the symmetric DTW and the simplified DTW combined with a multi-model multiway principal component analysis (MMPCA) are applied to the synchronization of the multivariate trajectories of an industrial batch process. The industrial application illustrates that the detection and diagnosis capabilities of the two new monitoring schemes are comparable to those achieved by primitive MMPCA and MPCA.