Multidimensional Design Space Exploration for Evolving Manufacturing Systems: Interplay of Deployment, Topology, and Production Planning Architectural Decisions

abstract

Optimizing evolving manufacturing systems often requires considering the interplay of various architectural design decisions (ADDs), especially at the lower levels of the automation pyramid. Here, decisions about programmable logic controllers, stations, and shop-floor topology directly impact product characteristics. This necessitates a multidimensional design space exploration (DSE) approach that synthesizes diverse ADDs together. However, integrating diverse ADDs within the complex context of industrial automation remains a challenge, and existing DSE solutions from other domains often lack direct applicability. This work introduces a novel multidimensional DSE approach for such systems, leveraging domain-specific input models to automatically generate valid and optimal deployment, shop-floor topology, and production planning ADDs. We evaluate the approach by assessing the quality of derived ADDs against various domain optimization goals. This evaluation compares our approach to both random and manual ADDs, utilizing a prototype implementation and realistic manufacturing architectures.