Study Of Batch Process Monitoring Based On Process Data

Modern industrial processes have become more and more large-scale and complex, so the product quality and process safety are two important issues in industries. Also, process monitoring and fault detection have attracted more and more attention. Therefore, it is very necessary and desirable to solve the problems that how to extract useful information from the large amounts of data, and to utilize the obtained information for process safety and product quality control. In this context situation, data-based multivariate statistical methods have become more popular and have been successfully applied in process modeling, monitoring and controlling.Traditional multivariate statistical process monitoring (MSPM) methods mainly contain principal component analysis (PCA) and partial least squares (PLS).Method like MPCA and MPLS have already been used in industry process as extension of traditional multivariate statistics method. However, the original MPCA/MPLS method has several limitations such like, have to predict future data, hard to deal with Uneven-length batch data. A novel method for uneven-length-stage batch process by using a new stage segmentation strategy based on k-means clustering method (KM-PCA) was proposed and then a method combine PCA and JITL strategy(JITL-PCA) have proposed for overcoming the disadvantage of KM-PCA. The effectiveness and utility of the proposed methods were validated through the simulation benchmark of fed-batch penicillin and semiconductor production.