Improved Phase-based Calibration Modelling and Quality Prediction by Investigating the Effects of Inter-phase Correlation

Phase-based quality analysis and prediction has been widely addressed by employing different calibration modeling techniques in multiphase batch processes. In this paper, a rational analysis scheme is presented to evaluate and understand the effects of the inter-phase correlation on, such as the extraction of the latent information, model structure and quality prediction. This is performed by combining partial least squares and principal component of predictions and implementing it bi-directionally (Bi-PLS-PCP). Within each phase, it separates the process systematic variation into the common and unique parts respectively based on their changes under the influence of the inter-phase correlation. They can then be quantitatively evaluated and made better use of for enhanced process understanding and improved quality prediction. The strength and efficiency of the proposed algorithm are verified on a typical multiphase batch process, injection molding.