A nonlinear modeling and online monitoring method for the batch process using multiple local PCA

Producing good quality products is the common objective in industries. However, Achieving this objective can be very difficult in batch process, especially when quality measurements are not available on-line or they have long time delays. This paper proposes a simple and straight nonlinear dynamic modeling with multi-model structure for batch process online monitoring using minimum window principal component analysis (MWPCA). MWPCA replaces single linear MPCA with multiple local linear sub-models for batch process modeling. It does not estimate any deviations of the ongoing batch from the average trajectories. Since the proposed method eliminates prediction error, the accuracy of process performance monitoring increases. MWPCA modeling procedures, principle and its application are discussed in detail. The presented methodology is successfully applied to PVC batch process.