Step-wise sequential phase partition algorithm and on-line monitoring strategy for multiphase batch processes

As batches operate at different statuses across different phases, it can be advantageous to partition the whole batch process into different phases and characterize them separately by multiple local phase models. The conventional clustering-based phase division algorithm overlooks the time sequence of batch operation which thus may mix different time segments located within a batch into one phase. Moreover, it is hard to capture the transitions between neighboring phases. In the present work, an automatic step-wise sequential phase division algorithm is developed to capture the changes of process characteristics along time direction within each batch. Its theoretical support is framed and the related statistical characteristics are analyzed. Using this algorithm, major phases are captured and the transition regions are separated from them as separate time regions. Thus, different statistical models are developed to reflect their time-varying characteristics. The online monitoring system is set up, which can realtime judge the affiliation of each new sample and check its status by adopting the proper statistical model. Comprehensive comparison is conducted between the proposed algorithm and clustering-based phase division algorithm. Its feasibility and performance are illustrated by an injection molding process which presents typical multiphase nature as well as transition characteristics.