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.