Adaptive moving window MPCA for online batch monitoring

Online monitoring of chemical process performance is extremely important to ensure plant safety and produce consistent high-quality products. Multivariate statistical process control has found wide application in performance analysis, monitoring and fault diagnosis of batch processes using existing rich historical database. In this paper, we propose a simple and straight multivariate statistical modeling based on an adaptive moving window MPCA algorithm for monitoring the progress of batch processes in real-time. The method replaces an invariant fixed-model monitoring approach with adaptive updating model data structure within batch-to-batch, which overcomes the problem of changing operation condition and slow timevarying behavior of industrial processes. Moving window MPCA algorithm within batch can copy seamlessly with variable run length and needn't estimate any deviations of the ongoing batch from the average trajectories. The presented method is successful applied to a polymerization reactor of industrial Polyvinyl chloride (PVC) batch process in the JinXi chemical plant.