TWO-DIMENSIONAL DYNAMIC PCA WITH AUTO-SELECTED SUPPORT REGION

Support region selection is a key step of two-dimensional dynamic PCA modeling for batch process dynamics, as it can affect the accuracy of the model and the efficiency of monitoring and fault diagnosis. In this paper, an automatic method for support region selection is developed. This data-based method can be applied universally on different batch processes without any prior process knowledge. Simulation shows that developed method has good application potentials for both monitoring and fault diagnosis.