Hierarchical data‐driven adaptive quality control for a pharmaceutical feeding‐blending unit subjected to external and internal disturbances

Abstract In the context of Pharma 4.0, continuous pharmaceutical manufacturing (CPM) has emerged as the unequivocal frontrunner, rendering its pharmaceutical quality control (PQC) an indispensable imperative task. The feeding‐blending unit (FBU) serves as the pivotal initial stage of CPM, and the collaborative control of active pharmaceutical ingredient (API) concentration and blender outlet mass‐flow‐rate (BOM) at the exit of FBU is critical in achieving pharmaceutical quality standards and ensuring the stability of CPM system. Regarding this problem, this article initially defines, at the theoretical level, the collaborative control problem related to API concentration and BOM for FBU, and puts forward a hierarchical framework for collaborative PQC based on a data ‐ driven approach. Within this framework, the upper‐level design incorporates an event‐triggered data‐driven adaptive iterative learning controller (ET‐DDAILC), which is supported by state‐feedback control, for the dynamic regulation of API concentration under external disturbances; Meanwhile, the lower‐level also integrates a DDAILC to effectively suppress or alleviate the rapidly changing BOM caused by the actions of the upper‐level controller (referred to as internal disturbance). Finally, the effectiveness and superiority of our proposed framework are validated through a group of comparative computational experiments. Meanwhile, our method exhibited superior performance in comparison to other baseline approaches.